Upload
others
View
22
Download
0
Embed Size (px)
Citation preview
3GPP2 C.R1002-A
Version 1.0
Date: May 11th, 2009
cdma2000 Evaluation Methodology
Revision A
COPYRIGHT 2009
3GPP2 and its Organizational Partners claim copyright in this document and individual
Organizational Partners may copyright and issue documents or standards publications in
individual Organizational Partner’s name based on this document. Requests for reproduction
of this document should be directed to the 3GPP2 Secretariat at [email protected].
Requests to reproduce individual Organizational Partner’s documents should be directed to
that Organizational Partner. See www.3gpp2.org for more information.
ii
This page intentionally left blank.
3GPP2 C.R1002-A v1.0
CONTENTS
i
FOREWORD ................................................................................................................... xiii 1
REFERENCES ................................................................................................................. xv 2
1 Introduction ................................................................................................................ 1 3
1.1 Study Objective and Scope ........................................................................................ 1 4
1.2 Simulation Description Overview .............................................................................. 1 5
2 Evaluation Methodology for the Forward Link .............................................................. 3 6
2.1 System Level Setup ................................................................................................... 3 7
2.1.1 Antenna Pattern ............................................................................................ 3 8
2.1.2 System Level Assumptions ............................................................................ 3 9
2.1.3 Dynamical Simulation of the Forward Link Overhead Channels ................... 10 10
2.1.4 Reverse Link Modeling in Forward Link System Simulation ......................... 11 11
2.1.5 Signaling Errors .......................................................................................... 11 12
2.1.6 Fairness Criteria .......................................................................................... 12 13
2.1.6.1 Fairness Criterion with the Normalized CDF of the User 14
Throughput ....................................................................................................... 12 15
2.1.6.1.1 A Generic Proportional Fair Scheduler .............................................. 14 16
2.1.6.2 Fairness Criterion with Geometric Mean and Harmonic Mean ............... 15 17
2.1.7 C/I Predictor Model for System Simulation .................................................. 16 18
2.2 Link Level Modeling ................................................................................................ 16 19
2.2.1 Link to System FER mapping ...................................................................... 16 20
2.2.1.1 Quasi-Static Approach with Fudge Factors: .......................................... 17 21
2.2.1.2 Quasi-Static Approach with Short Term FER: ....................................... 17 22
2.2.1.3 Equivalent SNR Approach: .................................................................... 19 23
2.2.2 Channel Models ........................................................................................... 19 24
2.2.2.1 Channels model based on ITU channel model ....................................... 19 25
2.2.2.2 Channels model based on SCM ............................................................. 21 26
2.2.2.2.1 Channel model for system level simulations ...................................... 21 27
2.2.2.2.2 Channel model for link level simulations ........................................... 22 28
2.2.2.2.3 Channel model for virtual decoder generation and verification ........... 23 29
2.2.3 C/I modeling for system simulation ............................................................. 23 30
2.3 Simulation Flow and Output Matrices ..................................................................... 27 31
2.3.1 Simulation Flow for the Center Cell Method ................................................. 27 32
3GPP2 C.R1002-A v1.0
CONTENTS
ii
2.3.2 Simulation Flow for the Iteration Method ..................................................... 29 1
2.3.3 Simulation Flow for the Wrap Around Method ............................................. 31 2
2.3.4 Layout Files ................................................................................................. 32 3
2.3.5 Output Matrices .......................................................................................... 33 4
2.3.5.1 General output matrices ....................................................................... 34 5
2.3.5.2 Data Services and Related Output Matrices .......................................... 35 6
2.3.5.3 1xEV-DV Systems Only ......................................................................... 37 7
2.3.5.3.1 Voice Services and Related Output Matrices ...................................... 37 8
2.3.5.3.2 Mixed Voice and Data Services .......................................................... 37 9
2.3.5.4 1xEV-DO Systems Only (Mixed Rev. 0 and Rev. A Mobiles) ................... 38 10
2.3.5.5 UMB Systems Only ............................................................................... 38 11
2.4 Calibration Requirements ....................................................................................... 38 12
2.4.1 Link Level Calibration .................................................................................. 38 13
2.4.2 System Level Calibration ............................................................................. 39 14
2.4.2.1 UMB System Calibration ....................................................................... 39 15
2.4.2.1.1 Scheduler .......................................................................................... 39 16
3 Evaluation Methodology for the Reverse Link ............................................................. 43 17
3.1 System Level Setup ................................................................................................. 43 18
3.1.1 Antenna Pattern .......................................................................................... 43 19
3.1.2 System Level Assumptions .......................................................................... 43 20
3.1.3 Call Setup Model ......................................................................................... 48 21
3.1.4 Packet Scheduler ......................................................................................... 49 22
3.1.5 Backhaul Overhead Modeling in Reverse Link System Simulations .............. 49 23
3.1.6 Simulation of Forward Link Overheads for Reverse Link System 24
Simulation ........................................................................................................... 50 25
3.1.6.1 Static Modeling Method......................................................................... 50 26
3.1.6.2 Dynamic Modeling Method .................................................................... 51 27
3.1.6.3 Quantification of Forward Link Overhead as a Data Rate Cost 28
(1xEV-DV Systems Only) ................................................................................... 52 29
3.1.7 Signaling Errors .......................................................................................... 52 30
3.1.8 Fairness Criteria .......................................................................................... 53 31
3.1.9 FER Criterion .............................................................................................. 53 32
3.1.10 IoT Criterion ................................................................................................ 53 33
3GPP2 C.R1002-A v1.0
CONTENTS
iii
3.2 Link Level Modeling ................................................................................................ 54 1
3.2.1 Link Level Parameters and Assumptions ..................................................... 54 2
3.2.1.1 Frame Erasures .................................................................................... 54 3
3.2.1.2 Target FER ............................................................................................ 55 4
3.2.1.3 Channel Models .................................................................................... 56 5
3.2.2 Forward Link Loading .................................................................................. 57 6
3.2.3 Reverse Link Power Control ......................................................................... 57 7
3.3 Simulation Requirements ........................................................................................ 58 8
3.3.1 Simulation Flow .......................................................................................... 58 9
3.3.1.1 Soft and Softer Handoff ......................................................................... 58 10
3.3.1.2 Simulation Description ......................................................................... 59 11
3.3.1.3 Layout Files .......................................................................................... 60 12
3.3.2 Outputs and Performance Metrics ............................................................... 61 13
3.3.2.1 General Output Matrices ...................................................................... 61 14
3.3.2.2 Data Services and Related Output Matrices .......................................... 61 15
3.3.2.3 1xEV-DV Systems Only ......................................................................... 63 16
3.3.2.3.1 Voice Services and Related Output Matrices ...................................... 63 17
3.3.2.3.2 Mixed Voice and Data Services .......................................................... 63 18
3.3.2.4 Mixed Rev. 0 and Rev. A Mobiles (1xEV-DO Systems Only) ................... 64 19
3.3.2.5 UMB Systems Only ............................................................................... 64 20
3.3.2.6 Link Level Output ................................................................................. 65 21
3.4 Calibration Requirements ....................................................................................... 65 22
3.4.1 Link Level Calibration .................................................................................. 65 23
3.4.2 System Level Calibration ............................................................................. 65 24
3.4.2.1 1xEV-DV System Calibration ................................................................ 65 25
3.4.2.2 1xEV-DO System Calibration ................................................................ 65 26
3.4.2.3 UMB System Calibration ....................................................................... 66 27
3.4.2.3.1 Scheduler .......................................................................................... 67 28
3.5 1xEV-DO Baseline Simulation Procedures .............................................................. 70 29
3.5.1 Access Terminal Requirements and Procedures: .......................................... 70 30
3.5.2 Access Network Requirements and Procedures: ........................................... 71 31
3.5.3 Simulation Procedures ................................................................................ 71 32
3GPP2 C.R1002-A v1.0
CONTENTS
iv
4 Traffic Service Models ................................................................................................ 74 1
4.1 Forward Link Services............................................................................................. 74 2
4.1.1 Service Mix (1xEV-DV Systems Only) ........................................................... 74 3
4.1.2 TCP Model ................................................................................................... 74 4
4.1.3 HTTP Model ................................................................................................. 80 5
4.1.3.1 HTTP Traffic Model Characteristics ....................................................... 80 6
4.1.3.2 HTTP Traffic Model Parameters ............................................................. 82 7
4.1.3.2.1 Packet Arrival Model for HTTP/1.0-Burst Mode ................................. 84 8
4.1.3.2.2 Packet Arrival Model for HTTP/1.1-Persistent Mode .......................... 86 9
4.1.4 FTP Model ................................................................................................... 89 10
4.1.4.1 FTP Traffic Model Characteristics .......................................................... 89 11
4.1.4.2 FTP Traffic Model Parameters................................................................ 89 12
4.1.5 WAP Model .................................................................................................. 91 13
4.1.6 Near Real Time Video Model ........................................................................ 92 14
4.1.7 Voice Model (1xEV-DV Systems Only) .......................................................... 94 15
4.1.8 Delay Criteria .............................................................................................. 94 16
4.1.8.1 Performance Criteria for Near Real Time Video ...................................... 95 17
4.1.8.2 Delay Criterion for WAP Users............................................................... 95 18
4.2 Reverse Link Services ............................................................................................. 95 19
4.2.1 Service Mix (1xEV-DV Systems Only) ........................................................... 95 20
4.2.1.1 Data Model ........................................................................................... 96 21
4.2.1.2 Traffic Model ......................................................................................... 96 22
4.2.2 TCP Modeling .............................................................................................. 97 23
4.2.3 FTP Upload / Email ................................................................................... 102 24
4.2.4 HTTP Model ............................................................................................... 103 25
4.2.4.1 HTTP Traffic Model Parameters ........................................................... 104 26
4.2.4.2 Packet Arrival Model for HTTP ............................................................. 106 27
4.2.4.3 Forward Link Delay Model for HTTP Users .......................................... 108 28
4.2.5 WAP Users ................................................................................................. 109 29
4.2.6 Reverse Link Delay Criteria for HTTP/WAP ................................................ 111 30
4.2.7 Mobile Network Gaming Model .................................................................. 112 31
4.2.8 Voice Model (1xEV-DV Systems Only) ........................................................ 113 32
3GPP2 C.R1002-A v1.0
CONTENTS
v
4.3 Common Traffic Models Applicable for Both Forward Link and Reverse Link 1
Services ................................................................................................................ 113 2
4.3.1 Voice over IP(VoIP) ..................................................................................... 113 3
4.3.1.1 Source Configuration Files .................................................................. 113 4
4.3.1.2 Source Files ........................................................................................ 113 5
4.3.1.3 Simulation Specifics ........................................................................... 114 6
4.3.1.4 VoIP Statistics ..................................................................................... 115 7
4.3.1.5 Source Mix .......................................................................................... 116 8
4.3.1.6 Scheduler Statistics ............................................................................ 116 9
4.3.2 Video Telephony(VT) .................................................................................. 116 10
4.3.2.1 Source Configuration Files .................................................................. 116 11
4.3.2.2 Source Files ........................................................................................ 116 12
4.3.2.3 Simulation Specifics ........................................................................... 117 13
4.3.2.4 VT Statistics ....................................................................................... 118 14
4.3.2.5 Source Mix .......................................................................................... 119 15
4.3.2.6 Scheduler Statistics ............................................................................ 119 16
Appendix A: Lognormal description ............................................................................... 121 17
Appendix B: Antenna Orientation ................................................................................. 122 18
Appendix C: Definition of System Outage and Voice Capacity ........................................ 124 19
Appendix D: Formula to define various throughput and Delay Definitions ..................... 125 20
Appendix E: Link budget ............................................................................................... 128 21
Appendix F: Quasi-static Method for Link Frame Erasures Generation and Dynamically 22
Simulated Forward Link Overhead Channels ................................................................. 135 23
Appendix G: Equalization .............................................................................................. 146 24
Appendix H: Max-Log-Map Turbo Decoder Metric .......................................................... 149 25
Appendix I: 19 Cell Wrap-Around Implementation ......................................................... 151 26
Appendix J: Link Level Simulation Parameters .............................................................. 155 27
Appendix K: Joint Technical Committee (JTC) Fader ..................................................... 159 28
Appendix L: Largest Extreme Value Distribution ........................................................... 163 29
Appendix M: Reverse Link Output Matrices ................................................................... 164 30
M.1 Output Matrix ..................................................................................................... 164 31
M.1.1 1xEV-DV Systems ............................................................................................. 164 32
M.1.2 1xEV-DO Systems ............................................................................................ 171 33
3GPP2 C.R1002-A v1.0
CONTENTS
vi
M.2 Definitions ........................................................................................................... 176 1
Appendix N: Link Prediction Methodology for Uplink System Simulations ...................... 179 2
N.1 Definition of Required Terms ................................................................................ 182 3
N.1.1 Channel Estimation SNR, ,i p ........................................................................ 182 4
N.1.2 Non-Gaussian Penalty, NG ............................................................................ 183 5
N.1.3 Reference Curves ............................................................................................ 183 6
Appendix O: Reverse Link Hybrid ARQ: Link Error Prediction Methodology Based on Convex 7
Metric ............................................................................................................................ 185 8
O.1 Convex Metric based on Channel Capacity Formula ............................................. 185 9
O.2 Equivalent SNR Method based on Convex Metric (ECM) ....................................... 188 10
O.2.1 Overview of the Procedure .............................................................................. 188 11
O.2.2 Detailed Procedure ......................................................................................... 189 12
O.2.3 Combining Procedure for H-ARQ .................................................................... 191 13
Appendix P: Pilot SINR Estimation For Power-Control Command Update in Link-Level 14
Simulations ................................................................................................................... 192 15
Appendix Q: 1xEV-DV Reverse Link Simulation and Scheduler Procedures ................... 193 16
Q.1.1 Mobile station Requirements and Procedures .................................................... 193 17
Q.1.2 Base Station Requirements and Procedures ...................................................... 196 18
Q.1.3 Scheduler Requirements and Procedures .......................................................... 198 19
Q.1.3.1 Scheduling, Rate Assignment and Transmission Timeline ........................... 199 20
Q.1.3.2 Scheduler Description and Procedures ........................................................ 201 21
Q.2 Baseline specific simulation parameters ............................................................... 204 22
Appendix R: Modeling of D_RL(request) and D_FL(Assign) ............................................. 207 23
Appendix S: Symbol SNR Modeling for CDM transmission with Rake demodulation ...... 213 24
Appendix T: Equivalent SNR Approach for OFDM Transmission and Demodulation ...... 216 25
T.1 Coherence Loss due to Doppler ............................................................................. 216 26
T.2 Inter-tone Interference (ITI) due to Doppler ........................................................... 216 27
T.3 Channel Estimation Loss and Pilot Weighted Combining (PWC) ............................ 216 28
T.4 Antenna Combining with Receive Diversity ........................................................... 217 29
T.5 Averaging SNR in the constrained capacity domain .............................................. 218 30
Appendix U: File formats for VoIP and VT model ............................................................ 220 31
U.1 Source Configuration File Format......................................................................... 220 32
3GPP2 C.R1002-A v1.0
CONTENTS
vii
U.2 Source File Format ............................................................................................... 220 1
U.3 Per-AT Data Reporting Format for VoIP ................................................................ 220 2
U.4 Network Statistics for VoIP ................................................................................... 221 3
U.5 Per-AT Data Reporting Format for VT ................................................................... 221 4
U.6 Network Statistics for VT ...................................................................................... 221 5
Appendix V: Channel parameters for furge factor .......................................................... 223 6
Appendix W: link level statistics for generating the short-term FER curves for link-to-7
system mapping ............................................................................................................ 235 8
W.1 Terminology ......................................................................................................... 235 9
W.2 PER and SINR Definitions .................................................................................... 236 10
W.3 Examples ............................................................................................................ 237 11
12
3GPP2 C.R1002-A v1.0
CONTENTS
viii
This page intentionally left blank 1
.2
3GPP2 C.R1002-A v1.0
FIGURES
ix
Figure 2.1.1-1 Antenna Pattern for 3-Sector Cells .............................................................. 3 1
Figure 2.3.1-1 Simulation Flow Chart .............................................................................. 29 2
Figure 3.1.3-1: Simplified Call Setup Timeline for 1xEV-DV. (Timeline for 1xEV-DO is the 3
same by modifying 320 ms to 427 ms) ....................................................................... 49 4
Figure 4.1.2-1 Control Segments in TCP Connection Set-up and Release.......................... 75 5
Figure 4.1.2-2 TCP Flow Control During Slow-Start; l = Transmission Time over the Access 6
Link; rt = Roundtrip Time .......................................................................................... 76 7
Figure 4.1.2-3 Packet Arrival Process at the Base Station for the Download of an Object 8
Using TCP; PW = the Size of the TCP Congestion Window at the End of Transfer of the 9
Object; Tc=c (Described in Figure 4.1.2-2) .................................................................. 79 10
Figure 4.1.3-1 Packet Trace of a Typical Web Browsing Session ....................................... 80 11
Figure 4.1.3-2 Contents in a Packet Call .......................................................................... 81 12
Figure 4.1.3-3 A Typical Web Page and Its Content .......................................................... 81 13
Figure 4.1.3-4 Modeling a Web Page Download ................................................................. 84 14
Figure 4.1.3-5 Download of an Object in HTTP/1.0-Burst Mode ....................................... 86 15
Figure 4.1.3-6 Download of Objects in HTTP/1.1-Persistent Mode .................................... 88 16
Figure 4.1.4-1 Packet Trace in a Typical FTP Session ....................................................... 89 17
Figure 4.1.4-2 Model for FTP Traffic ................................................................................. 91 18
Figure 4.1.5-1 Packet Trace for the WAP Traffic Model ...................................................... 92 19
Figure 4.1.6-1 Video Streaming Traffic Model ................................................................... 93 20
Figure 4.2.2-1: Modeling of TCP three-way handshake ..................................................... 98 21
Figure 4.2.2-2: TCP Flow Control During Slow-Start; l = Transmission Time over the 22
Access Link (RL); rt = Roundtrip Time ........................................................................ 99 23
Figure 4.2.2-3 Packet Arrival Process at the mobile Station for the Upload of a File Using 24
TCP .......................................................................................................................... 102 25
Figure 4.2.4-1: Example of events occurring during web browsing. ................................. 104 26
Figure 4.2.5-1: Packet Trace for the WAP Traffic Model ................................................... 110 27
Figure B-1 Center Cell Antenna Bearing Orientation diagram ......................................... 122 28
Figure B-2 Orientation of the Center Cell Hexagon ......................................................... 122 29
Figure B-3 Mobile Bearing orientation diagram example. ................................................ 123 30
Figure D-1: Description of arrival and delivered time for a packet and a packet call. ...... 127 31
Figure F-1 Flowchart for QPSK modulation .................................................................... 139 32
Figure F-2 Prediction methodology for higher order modulations without pure Chase 33
combining ................................................................................................................ 140 34
3GPP2 C.R1002-A v1.0
FIGURES
x
Figure F-3 Prediction methodology for higher order modulations with pure Chase 1
combining (this corresponds to Case 1) .................................................................... 141 2
Figure F-4 Obtaining sample values of
s tE /N and indicators of packet errors ............... 142 3
Figure F-5 Determining the Doppler penalty by evaluating the predictor performance .... 143 4
Figure H-1 QAM Receiver Block Diagram ........................................................................ 150 5
Figure I-1 Wrap-around with ‘9‘ sets of 19 cells showing the toroidal nature of the wrap-6
around surface. ........................................................................................................ 152 7
Figure I-2: The antenna orientations to be used in the wrap-around simulation. The arrows 8
in the Figure show the directions that the antennas are pointing. ............................ 154 9
Figure K-1: I and Q Fade Multiplier Generation .............................................................. 159 10
Figure N-1: Outline of Equivalent SNR Method. .............................................................. 179 11
Figure O-1. Approximate channel capacities at BPSK and QPSK signaling (Gaussian 12
signaling case is plotted as a reference). ................................................................... 187 13
Figure Q-1: Set point adjustment due to rate transitions on R-SCH ................................ 198 14
Figure Q-2 Scheduling Delay Timing .............................................................................. 199 15
Figure Q-3: Parameters associated in mobile station scheduling on RL ........................... 200 16
Figure R-1: PDF of FL transmission delays of ESCAM on F-PDCH .................................. 210 17
Figure R-2: ESCAM delays on F-PDCH ........................................................................... 211 18
Figure T-1 Constrained Capacity Curve for 16-QAM ....................................................... 219 19
20
3GPP2 C.R1002-A v1.0
TABLES
xi
Table 2.1.2-1 Forward System Level Simulation Parameters ............................................... 4 1
Table 2.1.2-2 Details of Self-Interference Values Resulting in 13.5 dB of Maximum C/I for 2
CDM Transmission with Rake Demodulation ............................................................... 8 3
Table 2.1.2-3 Details of Self-Interference Values Resulting in 17.8 dB of Maximum C/I for 4
CDM Transmission with Rake Demodulation ............................................................... 8 5
Table 2.1.2-4 Details of Self-Interference Values Resulting in 17 dB of Maximum C/I for 6
OFDM Transmission and Demodulation ....................................................................... 9 7
Table 2.1.5-1 Signaling Errors .......................................................................................... 12 8
Table 2.1.6-1 Criterion CDF ............................................................................................. 13 9
Table 2.1.6-2 Web Browsing Model Parameters ................................................................ 14 10
Table 2.2.2-1 Channel Models .......................................................................................... 19 11
Table 2.2.2-2 Fractional Recovered Power and Fractional UnRecovered Power .................. 20 12
Table 2.2.2-3 Relative Power of each Multipath Model (in dB) ........................................... 20 13
Table 2.2.2-4 Delay of each Multipath Model (in ns) ........................................................ 20 14
Table 2.3.5-1 Required 1xEV-DV Simulation Evaluation Comparison Cases Table ........... 34 15
Table 3.1.2-1 Reverse Link System Level Simulation Parameters ...................................... 44 16
Table 3.1.5-1 Backhaul bandwidth used by signaling and measurement messages .......... 50 17
Table 3.1.8-1 CDF Criterion for FTP Upload MS ............................................................... 53 18
Table 3.4.2-1 Default 1xEV-DO RL MAC Transition Probabilities ...................................... 66 19
Table 4.1.3-1 HTTP Traffic Model Parameters ................................................................... 83 20
Table 4.1.4-1 FTP Traffic Model Parameters ...................................................................... 90 21
Table 4.1.5-1 WAP Traffic Model Parameters .................................................................... 92 22
Table 4.1.6-1 Video Streaming Traffic Model Parameters .................................................. 94 23
Table 4.2.1-1: Traffic Configurations ................................................................................ 96 24
Table 4.2.2-1 Delay components in the TCP model for the RL upload traffic ................... 100 25
Table 4.2.3-1: FTP Characteristics .................................................................................. 103 26
Table 4.2.4-1: HTTP Traffic Model Parameters ................................................................ 106 27
Table 4.2.4-2 Points to obtain the average transmission rate (ATR) given the geometry and 28
channel model of a user ........................................................................................... 109 29
Table 4.2.5-1: WAP Traffic Model Parameters ................................................................. 111 30
Table 4.2.6-1 Reverse link delay criteria for HTTP request .............................................. 111 31
Table 4.2.7-1 Mobile network gaming traffic model parameters ...................................... 112 32
Table E-1 Link-Budget Template for the Reverse Link..................................................... 129 33
3GPP2 C.R1002-A v1.0
TABLES
xii
Table E-2 Link-Budget Template for the Forward Link .................................................... 131 1
Table E-3 Propagation Index and Log-Normal Sigma Values from [20] ............................ 134 2
Table J-1 Link Level Simulation Parameters for Forward Link ........................................ 155 3
Table J-2 Link Level Simulation Parameters for Reverse Link ......................................... 157 4
Table K-1: Coefficients of the 6-tap FIR Filter ................................................................. 159 5
Table K-2: Coefficients of the 8-tap FIR Filter ................................................................. 159 6
Table K-3: Coefficients of the 11-tap Filter ...................................................................... 160 7
Table K-4: Coefficients of the 28-tap Filter ...................................................................... 160 8
Table K-5: Jakes (Classic) Spectrum IIR Filter Coefficients ............................................. 161 9
Table M-1 Required statistics output in excel spread sheet for the base station side ....... 164 10
Table M-2 Required statistics output in excel spread sheet for the base station side ....... 170 11
Table M-3 Required statistics output in excel spread sheet for the base station side ....... 171 12
Table N-1. Notations used. ............................................................................................. 180 13
Table R-1: D_RL(request) delay for Method a .................................................................. 207 14
Table R-2: D_RL(request) delay for Method b .................................................................. 208 15
Table R-3: D_FL(assign) delay for Method a .................................................................... 208 16
Table R-4: D_FL(assign) delay for Method b (excluding F-PDCH scheduling delay) .......... 209 17
Table R-5: Reference table for mean transmission times vs Geometry ............................. 211 18
19
3GPP2 C.R1002-A v1.0
FOREWORD
xiii
This document was prepared by Technical Specification Group C of the Third Generation 3 1
Partnership Project 2 (3GPP2). 2
Revision 0 of this document was used for the evaluation and analysis leading to the 3
development of the following cdma2000 systems specifications: cdma2000®1 Revision C 4
(1xEV-DV), cdma2000 Revision D (1xEV-DV), and cdma2000 High Rate Packet Data Air 5
Interface Revision A (1xEV-DO). 6
Revision A of this document includes evaluation methodology for cdma2000 High Rate 7
Broadcast-Multicast Packet Data Air Interface (1xEV-DO BCMCS) and UMB (Ultra Mobile 8
Broadband)2. 9
10
1 cdma2000® is the trademark for the technical nomenclature for certain specifications and
standards of the Organizational Partners (OPs) of 3GPP2. Geographically (and as of the date of
publication), cdma2000® is a registered trademark of the Telecommunications Industry Association
(TIA-USA) in the United States.
2 Ultra Mobile Broadband™ and (UMB™) are trade and service marks owned by the CDMA
Development Group (CDG).
3GPP2 C.R1002-A v1.0
FOREWORD
xiv
This page intentionally left blank. 1
2
3GPP2 C.R1002-A v1.0
REFERENCES
xv
Normative References 1
The following specifications contain provisions which, through reference in this text, 2
constitute provisions of this Specification. At the time of publication, the editions indicated 3
were valid. If the specification version number is included, the reference is specific. Parties 4
implementing this Specification should use the specific versions of the indicated 5
specification. If the specification version number is not included, the reference is non-6
specific. Parties implementing this Specification are encouraged to investigate the 7
possibility of applying the most recent editions of the indicated specifications. 8
[1] ETSI TR 101 12, Universal Mobile Telecommunications System (UMTS); Selection 9
procedures for the choice of radio transmission technologies of the UMTS (UMTS 30.03 10
v3.2.0) 11
[2] ITU-RM.1225, Guidelines for Evaluation of Radio Transmission Technologies for IMT-12
2000. 13
[3] A. Viterbi, Principles of Spread Spectrum Communication, Addison-Wesley, 1995. 14
[4] F. Ling, ―Optimal Reception, performance bound, and cutoff rate analysis of reference-15
assisted coherent CDMA communications with applications,‖ IEEE Trans. on Commun., 16
46(10), pp. 1583—1592, October 1999. 17
[5] Motorola, "HTTP Traffic Models for 1xEV-DV Simulations", Contribution 3GPP2-C50-18
Eval-20010212-004. 19
[6] Motorola, " HTTP Traffic Models for 1xEV-DV Simulations (v2)", Contribution 3GPP2-20
C50-Eval-20010321-002-HTTP-traffic. 21
[7] R. Fielding, J. Gettys, J. C. Mogul, H. Frystik, L. Masinter, P. Leach, and T. Berbers-Lee, 22
"Hypertext Transfer Protocol - HTTP/1.1", RFC 2616, HTTP Working Group, June 1999. 23
ftp://ftp.Ietf.org/rfc2616.txt. 24
[8] B. Krishnamurthy and M. Arlitt, "PRO-COW: Protocol Compliance on the Web", 25
Technical Report 990803-05-TM, AT&T Labs, August 1999, 26
http://www.research.att.com/~bala/papers/procow-1.ps.gz. 27
[9] Lucent, "Comments on HTTP traffic model", Contribution 3GPP2-C50-Eval-20010323-28
001-traffic-comments. 29
[10] J. Cao, William S. Cleveland, Dong Lin, Don X. Sun., "On the Nonstationarity of 30
Internet Traffic", Proc. ACM SIGMETRICS 2001, pp. 102-112, 2001. 31
[11] B. Krishnamurthy, C. E. Wills, "Analyzing Factors That Influence End-to-End Web 32
Performance", http://www9.org/w9cdrom/371/371.html 33
[12] H. K. Choi, J. O. Limb, "A Behavioral Model of Web Traffic", Proceedings of the seventh 34
International Conference on Network Protocols, 1999 (ICNP '99), pages 327-334. 35
[13] F. D. Smith, F. H. Campos, K. Jeffay, D. Ott, "What TCP/IP Protocol Headers Can Tell 36
Us About the Web", Proc. 2001 ACM SIGMETRICS International Conference on 37
Measurement and Modeling of Computer Systems, pp. 245-256, Cambridge, MA June 38
2001. 39
3GPP2 C.R1002-A v1.0
REFERENCES
xvi
[14] P. Barford and M Crovella, "Generating Representative Web Workloads for Network and 1
Server Performance Evaluation" In Proc. ACM SIGMETRICS International Conference on 2
Measurement and Modeling of Computer Systems, pp. 151-160, July 1998. 3
[15] S. Deng. ―Empirical Model of WWW Document Arrivals at Access Link.‖ In Proceedings 4
of the 1996 IEEE International Conference on Communication, June 1996 5
[16] W. R. Stevens, "TCP/IP Illustrated, Vol. 1", Addison-Wesley Professional Computing 6
Series, 1994. 7
[17] Motorola, "Comments on Data Traffic Mix", Contribution 3GPP2-C50-Eval-20010321-8
006-Mot-traffic-mix. 9
[18] UMTS 30.03 V3.2.0 "Universal Mobile Telecommunications Systems (UMTS); Selection 10
procedures for the choice of radio transmission technologies of the UMTS, 1998-04, pg 33-11
35. 12
[19] K. C. Claffy, "Internet measurement and data analysis: passive and active 13
measurement", http://www.caida.org/outreach/papers/Nae/4hansen.html. 14
[20] ITU, ―Guidelines for Evaluation of Radio Transmission Technologies for IMT-2000,‖ 15
Recommendation ITU-R M.1225, 1997. 16
[21] Bob Love and Frank Zhou, ―Comments on Qualcomm Link Budget for 1xEV-DV,‖ 17
Motorola contribution 3GPP2-C50-Eval-20001205-001 to the Evaluation Ad Hoc, December 18
5, 2000. 19
[22] Tao Chen, ―Link Budget Examples for 1xEV-DV Proposal Evaluation, Rev. 2,‖ 20
QUALCOMM contribution 3GPP2-C50-Criteria Ad Hoc-20001115-002 to the WG5 Criteria 21
Ad Hoc, November 15, 2000. 22
[23] Louay Jalloul, ―Comments on Path Loss Models for System Simulations,‖ Motorola 23
contribution 3GPP2-C50-WG5-20001116-003 to the Simulation Ad Hoc, November 16, 24
2000. 25
[24] Steve Dennett, ―The cdma2000 ITU-R RTT Candidate Submission (0.18),‖ July 27, 26
1998 27
[25] Working Group 5 evaluation ad hoc chair, ―1xEV-DV Evaluation Methodology 28
Addendum, version 6‖, 3GPP2 TSG-C contribution to Working Group 5 in the Portland, 29
Oregon meeting, C50-20010820-026, August 20, 2001. 30
[26] 3GPP2/TSG-C - C50-Eval-20010329-001, ―Link Error Prediction Methodology,‖ Lucent 31
Technologies, March 2001. 32
[27] 3GPP2/TSG-C – C30-20030217-010, ―Link Prediction Methodology for Reverse Link 33
System Simulations,‖ Lucent Technologies, February 2003. 34
[28] 3GPP2/TSG-C – C30-20030217-010A ―Link Prediction Methodology for Uplink System 35
Level Simulations – Analysis,‖ Lucent Technologies, February 2003. 36
[29] TIA/EIA-IS-2000.2, ―Mobile Station-Base Station Compatibility Standard for Dual-37
Mode Wideband Spread Spectrum Cellular System, June, 2002. 38
3GPP2 C.R1002-A v1.0
REFERENCES
xvii
[30] Ramakrishna, S., Holtzman, J.M., ―A scheme for throughput maximization in a dual-1
class CDMA system ―, Selected Areas in Communications, IEEE Journal on, Vol. 16 Issue: 2
6, page 830 –844, Aug. 1998. 3
[31] F. Kelly, ―Charging and rate control for elastic traffic‖, European Trans. On 4
Telecommunications, vol. 8, pp. 33-37, 1997. 5
[32] L3NQS, ―Results of L3NQS Simulation Study‖, 3GPP2-C50-20010820-011, August 6
2001. 7
[33] Doug Reed, ―‘Modified‘ Hata path loss model used in 3GPP2 ―, 3GPP2 Contribution 8
C30-20040920-012, Motorola. 9
[34] 3GPP2/TSG-C – C30-20040823-063R1 Qualcomm, ―Confidence Interval‖, August 23rd, 10
2004. 11
[35] Ye Li, Leonard Cimini, ―Bounds on the Interchannel Interference of OFDM in Time-12
Varying Impairments‖, IEEE Transactions on Communications, Vol 49, No. 3, March 2001. 13
[36] 3GPP2/TSG-C WG3 contribution, C30-20060413-006, ―SampleSrcConfigFile_VTMix6 14
AT‖, April 2006. 15
[37] 3GPP2/TSG-C WG3 contribution, C30-20060413-005, ―MSOWithBlankingSourceFile‖, 16
April 2006. 17
[38] 3GPP2/TSG-C WG3 contribution, C30-20060327-030A, ―FixedRateFixedQualityVideo 18
SourceFile‖, April 2006. 19
[39] 3GPP2 C.S0076-0 v1.0, ―Discontinuous Transmission (DTX) of Speech in cdma2000 20
Systems‖, December, 2005. 21
[40] 3GPP2/TSG-C WG1 contribution, C12-20051012-006, ―Video Database for 3GPP2 22
multimedia services‖, October 2005. 23
[41] 3GPP2/TSG-C WG3 contribution, C30-20030915-006, ―SCM-135 Channel Model Text‖, 24
September 2006. 25
[42] 3GPP2/TSG-C WG3 contribution, C30-20060823-005, ―FL VoIP packet arrival with 26
jitter.dat,‖, August 2006. 27
[43] 3GPP2/TSG-C WG3 contribution, C30-20080114-030, ―Updated location files for 28
calibration,‖ January 2008. 29
[44] 3GPP2/TSG-C WG3 contribution, C30-20080114-029, ―Effective SNR AWGN Curves,‖ 30
January 2008. 31
[45] 3GPP2/TSG-C WG3 contribution, C30-20080331-012R3, ―Performance Evaluation 32
Parameters,‖ March 2008. 33
[46] 3GPP2/TSG-C WG3 contribution, C30-20080114-021R2, ―Calibration output metrics,‖ 34
January 2008. 35
[47] 3GPP2/TSG-C WG3 contribution, C30-20071203-020R2, ―SNR to CQI mapping in 36
calibration,‖ December 2007. 37
Informative References 38
3GPP2 C.R1002-A v1.0
REFERENCES
xviii
The following documents do not contain provisions of the Specification. They are listed to 1
aid in better understanding this Specification. 2
3
4
3GPP2 C.R1002-A v1.0
1
1 INTRODUCTION 1
1.1 Study Objective and Scope 2
The objective of this document is to explain the set of definitions, assumptions, and a 3
general framework for simulating cdma2000® systems (e.g., 1xEV-DV and 1xEV-DO) and 4
UMB® (Ultra Mobile Broadband) systems to arrive at system wide voice, data, or both voice 5
and data performance on the forward and reverse links. 6
This document was used in the evaluation and analysis leading to the development of the 7
following specifications: cdma2000 Revision C (1xEV-DV), cdma2000 Revision D (1xEV-DV), 8
cdma2000 High Rate Packet Data Air Interface Revision A (1xEV-DO), cdma2000 High Rate 9
Broadcast-Multicast Packet Data Air Interface (1xEV-DO BCMCS) and UMB. 10
This document also defines the necessary framework for simulating the performance of 11
cdma2000 and UMB systems with proposed enhancements that are not part of the current 12
cdma2000 and UMB family of specifications. The proponent(s) of any proposal shall provide 13
the details required so that other companies can evaluate the proposal independently. The 14
proponent(s) of any simulation results shall provide the details required so that other 15
companies can repeat the simulation independently. The information about the simulations 16
will include the predictors being used, and the reported results will include the prediction 17
errors (bias and standard deviation). 18
1.2 Simulation Description Overview 19
Determining voice and high rate packet data system performance requires a dynamic 20
system simulation tool to accurately model feedback loops, signal latency, protocol 21
execution, and random packet arrival in a multipath-fading environment. The packet 22
system simulation tool will include Rayleigh and Rician fading and evolve in time with 23
discrete steps (e.g., time steps of 1.25 ms or 1.67 ms). The time steps need to be small 24
enough to correctly model feedback loops, latencies, scheduling activities, and 25
measurements of the proposed system. 26
27
3GPP2 C.R1002-A v1.0
2
This page intentionally left blank. 1
3GPP2 C.R1002-A v1.0
3
2 EVALUATION METHODOLOGY FOR THE FORWARD LINK 1
2.1 System Level Setup 2
2.1.1 Antenna Pattern 3
The antenna pattern used for each sector, reverse link and forward link, is plotted in Figure 4
2.1.1-1 and is specified by 5
2
3
min 12 , , where 180 180.m
dB
A A
2.1-1 6
dB3 is the 3 dB beamwidth, and dBAm 20 is the maximum attenuation. 7
-25
-20
-15
-10
-5
0
-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180
Horizontal Angle - Degrees
Ga
in -
dB
8
Figure 2.1.1-1 Antenna Pattern for 3-Sector Cells 9
2.1.2 System Level Assumptions 10
The parameters used in the simulation are listed in Table 2.1.2-1. Where values are not 11
shown, the values and assumptions used shall be specified in the simulation description. 12
3GPP2 C.R1002-A v1.0
4
Table 2.1.2-1 Forward System Level Simulation Parameters 1
Parameter Value Comments
Number of Cells (3 sectored) 19 2 rings, 3-sector system, 57
sectors.
Antenna Horizontal Pattern 70 deg (-3 dB)
with 20 dB front-to-back
ratio
See section 2.1.1
Antenna Orientation 0 degree horizontal azimuth
is East (main lobe)
No loss is assumed on the
vertical azimuth. (See
Appendix B)
Propagation Model
(BTS Ant Ht=32m, MS=1.5m)
28.6+ 35log10(d) dB,
d in meters
Modified Hata Urban Prop.
Model @1.9GHz (COST 231).
Minimum of 35 meters
separation between MS and
BS.3
Log-Normal Shadowing Standard Deviation = 8.9 dB Independently generate
lognormal per mobile and
use the method described in
Appendix A. This shadowing
is constant for each MS in
each simulation run. The
same shadowing amount
shall be used for all the
sector antennas of a BS to a
given MS. The correlation
coefficient between the BS‘s
Tx antennas and a given MS
and the BS‘s RX antennas
and a given MS is 1.
Base Station Shadowing
Correlation
0.5 See Appendix A
3 In this document the word ―modified‖ represents a difference from the COST231-Hata model
wherein the path loss has been reduced by 3 dB [33]. If a mobile is dropped within 35 meters of a
base station, it shall be redropped until it is outside the 35-meter circle.
3GPP2 C.R1002-A v1.0
5
Parameter Value Comments
Forward Link
Overhead
Channel
Resource
Consumption
Circuit
switched
and packet
switched
data
systems
(e.g., 1xEV-
DV)
Pilot, Paging and Sync
overhead: 20%.
Any additional overhead
needed to support other
control channels (dedicated
or common) must be
specified and accounted for
in the simulation
Packet
switched
data
systems
(e.g., 1xEV-
DO)
FL MAC, Preamble and Pilot
channel overhead shall be
considered. The portion of
times the Control Channel
(CC) (38.4 kbps or 76.8
kbps) is sent shall be set as
a fixed TDM overhead.
CC portion is assumed to be
6.25% of the total time. Any
additional overhead must be
specified and accounted for
in the simulation.
Mobile Noise Figure 10.0 dB
Thermal Noise Density -174 dBm/Hz
Carrier Frequency 2 GHz
BS Antenna Gain with Cable
Loss
15 dB 17 dB BS antenna gain; 2
dB cable loss
MS Antenna Gain -1 dBi
Other Losses 10 dB Applicable to all fading
models
Fast Fading Model Based on Speed See Table 2.2.2-1. The fading
model is specified in
Appendix K. With dual
antenna receiver, the fading
processes on the paths from
a given BS to the MS receive
antennas are mutually
independent.
Active Set
Membership
Circuit
switched and
packet
switched
data systems
(e.g., 1xEV-
DV)
Up to 3 members are in the
Active Set if the pilot Ec/Io
is larger than T_ADD = -18
dB (=9 dB below the FL pilot
Ec/Ior) based on the FL
evaluation methodology
3GPP2 C.R1002-A v1.0
6
Parameter Value Comments
Packet
switched
data systems
(e.g., 1xEV-
DO)
Up to 3 members are in the
Active Set if the pilot Ec/Io
is larger than T_ADD = -9 dB
based on the FL evaluation
methodology
Finger Assignment
Threshold (T_PATH)
-12 dB A finger may be assigned to
a multipath component only
if its (Ec/Io) exceeds the
finger assignment threshold.
This parameter should be
used only for 1xEV-DO
BCMCS.
(See Appendix S)
Maximum Number of Paths
assigned to Rake fingers
(MAX_NUM_PATHS)
8 for single RX-antenna
receivers, and
4 for dual RX-antenna
receivers
This is the maximum
number of paths to which
Rake fingers may be
assigned at the MS.
Delay Spread Model See Table 2.2.2-1 and Table
2.2.2-2
Fast Cell Site Selection Disable. The overhead shall
be accounted for if it is used
in the proposal.
Forward Link
Power
Control
Circuit
switched and
packet
switched
data systems
(e.g., 1xEV-
DV)
(If used on
dedicated
channel)
Power Control loop delay:
two PCGs4
Update Rate: Up to 800Hz
PC BER: 4%
4 One PCG/slot delay in link level modeling (measured from the time that the SIR is sampled to the
time that the BS changes TX power level.)
3GPP2 C.R1002-A v1.0
7
Parameter Value Comments
Packet
switched
data systems
(e.g., 1xEV-
DO)
(If used on
MAC
channels)
Power Control loop delay:
two slots4
Update Rate: Up to 600 Hz
PC BER: 4%,
Based on DRC feedback
BS Maximum PA Power 20 Watts
Site to Site distance 2.5 km
1.9 km Determined by RL Link
Budget of 1xEV-D0 Rev-A
2.0 km Default site to site distance
for UMB
Total Path Loss Threshold 140 dB This term includes the MS
and BS antenna gains, cable
and connector losses, other
losses, and shadowing, but
not fading. A subscriber
whose total path loss on the
best forward link exceeds the
Total Path Loss Threshold is
redropped. This value
should be applied when site-
to-site distance is 1.9km and
2.0km.
Maximum C/I achievable,
where C is the
instantaneous total received
signal from the serving base
station(s) (usually also
referred to as rx_Ior(t), or
Îor(t)), and I is the
instantaneous total
interference level (usually
13.5 dB and 17.8 dB for
CDM transmission with
Rake demodulation
17 dB 18.1 dB5, and 28dB6
for OFDM transmission and
demodulation
13.5 dB for typical current
subscriber designs for IS-95
and cdma2000 1x systems;
17.8 dB for improved
subscriber designs for 1xEV-
DV and 1xEV-DO systems;
18.1dB and 28dB for
improved subscriber designs
for UMB. The details on how
these values were derived
5 The max AN C/I of 18.5 dB and the max AT C/I of 29dB are assumed. 18.1dB is the geometric
mean of those two values.
6 C/I required to meet 1% FER for the packet of 64QAM, code rate 3/4, and 2x2 SCW in channel of
PedB 3km/h is 26dB. 28dB max C/I is obtained by adding 2 dB margin.
3GPP2 C.R1002-A v1.0
8
Parameter Value Comments
also referred to as Nt(t)). are in Table 2.1.2-2, Table
2.1.2-3, and Table 2.1.2-4. (C/I)max for other
transmission/demodulation
schemes shall be provided
with justification, by the
system proponents.
1
Table 2.1.2-2 Details of Self-Interference Values Resulting in 13.5 dB of Maximum C/I 2
for CDM Transmission with Rake Demodulation 3
Contribution of Self-Interference )(ˆ/ˆ i
selfor II
Note
Base-band pulse shaping waveform 16.5dB IS-95 Tx filter and matched
Rx filter
Radio noise floor 20dB With improved Tx-Rx. Noise
performance
ADC quantization noise 20dB 4-bit A/D converter
Adjacent channel interference 27dB 1.25 MHz spacing
4
Table 2.1.2-3 Details of Self-Interference Values Resulting in 17.8 dB of Maximum C/I 5
for CDM Transmission with Rake Demodulation 6
Contribution of Self-Interference )(ˆ/ˆ i
selfor II Note
Base-band pulse shaping waveform 24 dB IS-95 Tx filter with 64-tap
Rx filter
Radio noise floor 20 dB For Tx RHO increased to
99%
ADC quantization noise 31.9 dB 6-bit A/D converter
Adjacent channel interference 27 dB 1.25 MHz spacing
7
8
9
10
3GPP2 C.R1002-A v1.0
9
Table 2.1.2-4 Details of Self-Interference Values Resulting in 17 dB of Maximum C/I 1
for OFDM Transmission and Demodulation 2
Contribution of Self-Interference )(ˆ/ˆ i
selfor II Note
Base-band pulse shaping waveform Not Applicable
OFDM transmission and
demodulation to eliminate
pulse ISI
Radio noise floor 20dB With improved Tx-Rx. Noise
performance
ADC quantization noise 20 dB 4-bit A/D converter
Adjacent channel interference Not applicable Guard tones to eliminate
ACI
3
The maximum C/I achievable in the subscriber receiver is limited by several sources, 4
including inter-chip interference induced by the base-band pulse shaping waveform, the 5
radio noise floor, ADC quantization error, and adjacent carrier interference. For 1xEV-DO 6
BCMCS, the noise floor associated with the maximum C/I limitation is modeled as 7
described in Section 2.2.2 and Appendix S. 8
In the system level simulation, the noise floor associated with the maximum C/I limitation 9
can be characterized by the parameter , given by 10
max/
1
IC 2.1-2 11
where max
IC denotes the maximum achievable C/I for the subscriber receiver. As 12
indicated in Table 2.1.2-1, max
IC is assumed to be 13.5 dB for the current IS-95 and 13
cdma2000 1X subscriber receivers, and 17.8 dB for improved 1xEV-DV/1xEV-DO designs. 14
Thus, = 0.045 and = 0.0166 for maximum C/I values of 13.5 dB and 17.8 dB, 15
respectively. 16
In the system level simulation for CDM transmission with Rake demodulation, the effective 17
C/I shall be given by 18
combined
effective
)/(
1
1
IC
IC 2.1-3 19
where combined
IC is the instantaneous signal-to-interference ratio after pilot-weighted 20
combining of the Rake fingers (see section 2.2.2 for detail). The effective signal-to-21
interference ratio, effective
IC , accounts for the interference sources associated with the 22
maximum C/I limitation, and shall be used as the C/I observed by the mobile station 23
receiver. 24
3GPP2 C.R1002-A v1.0
10
The channel between the serving cell and the subscriber is modeled using the channel 1
models defined in section 2.2.2. The channel between any interfering cell and the 2
subscriber can be modeled as a one-path Rayleigh fading channel, where the Doppler of the 3
fading process is randomly chosen based on the velocities specified in Table 2.2.2-1 and its 4
corresponding probabilities. 5
If transmit diversity is used in cdma2000 1x and 1xEV-DV systems, the transmit diversity 6
PA size shall be the same as the main PA size, 12.5% of the main PA power shall be used 7
for the Pilot Channel and 7.5% for the Paging Channel and Sync Channel. The Transmit 8
Diversity Pilot Channel power is half the power of the Pilot Channel. For example, if the 9
main PA size is 20 W, then the transmit diversity PA size is 20 W, 2.5 W of the main PA is 10
for the Pilot Channel, 1.5 W of the main PA for the Paging Channel and Sync Channel, and 11
1.25 W of the transmit diversity PA is for the Transmit Diversity Pilot Channel. 12
2.1.3 Dynamical Simulation of the Forward Link Overhead Channels 13
Dynamically simulating the overhead channels for 1xEV-DV or 1xEV-DO systems is 14
essential to capture the dynamic nature of power and code space allocation to these 15
channels. The simulations shall be done as follows: 16
1) The performance of the new overhead channels (other than the Pilot, Sync, and 17
Paging Channels for 1xEV-DV systems or the Pilot and control channels for 1xEV-18
DO systems) must be included in the system level simulations. The Pilot Channel, 19
Sync Channel, and Paging Channel are taken into account as part of the fixed 20
overhead (power and code space) in 1xEV-DV systems. For 1xEV-DO systems, the 21
Pilot, preamble, and the total FL MAC shall be transmitted at full BTS power (20 W), 22
and the 38.4 kbps and 76.8 kbps Control Channels are taken into account as part 23
of the fixed overhead (as a fixed percentage of the total transmission time). 24
2) There are two types of these new overhead channels: static and dynamic. A static 25
overhead channel requires fixed base station power. A dynamic overhead channel 26
requires dynamic base station power. 27
3) The system level simulations do not directly include the coding and decoding of 28
these new overhead channels. There are two aspects that are important for the 29
system level simulation: the required Ec/Ior during the simulation interval (e.g., a 30
power control group or slot) and demodulation performance (detection, miss, and 31
error probability — whatever is appropriate). 32
4) The link level performance is evaluated off-line by using separate link-level 33
simulations. A quasi-static approach shall be used to conduct the link-level 34
simulation. The performance is characterized by curves of detection, miss, false 35
alarm, and error probability (whatever is appropriate) versus Eb/No. 36
5) For static overhead channels, the system simulation should compute the received 37
Eb/No. 38
6) For dynamic overhead channels with open-loop control only, the simulations should 39
take into account the estimate of the required forward link power that needed to be 40
transmitted to the mobile station. For dynamic overhead channels that use closed 41
3GPP2 C.R1002-A v1.0
11
loop feedback, the base station allocates forward link power based upon the 1
combination of open-loop and closed-loop feedback. During the reception of 2
overhead information, the system simulation should compute the received Eb/No. 3
7) Once the received Eb/No is obtained, then the various miss error events should be 4
determined. The impact of these events should then be modeled. The false alarm 5
events are evaluated in link-level simulation, and the simulation results will be 6
included in the evaluation report. The impact of false alarm, such as delay increases 7
and throughput reductions for both the forward and reverse links, will be 8
appropriately taken into account in system-level simulation. 9
8) The Walsh space utilization shall be modeling dynamically for 1xEV-DV systems. 10
9) All new overhead channels shall be modeled. 11
10) If a proposal adds messages to an existing channel (overhead or otherwise), the 12
proponent shall justify that this can be done without creating undue loading on this 13
channel. If a proposal requires an additional overhead channel of the type that is 14
already in the system under evaluation, then the proposal shall include the power 15
required for this channel. The system level and link level simulation required for 16
this modified overhead channel as a result of the new messages shall be performed 17
according to 3) and 4), respectively. 18
2.1.4 Reverse Link Modeling in Forward Link System Simulation 19
The proponents shall only model feedback errors (e.g., power control, acknowledgements, 20
rate indication, etc.) and measurements (e.g., C/I measurement) without explicitly modeling 21
the reverse link and reverse link channels. In addition to supplying the feedback error rate 22
average and distribution, the measurement error model and selected parameters, the 23
estimated power level required for the physical reverse link channels will be supplied 24
(including those used for fast cell selection even though it is not going to be explicitly 25
modeled for the 1xEV-DV or 1xEV-DO system simulations). 26
2.1.5 Signaling Errors 27
Signaling errors shall be modeled and specified as in Table 2.1.5-1. 28
3GPP2 C.R1002-A v1.0
12
Table 2.1.5-1 Signaling Errors 1
Signaling Channel Errors Impact
ACK/NACK channel Misinterpretation, missed
detection, or false detection
of the ACK/NACK message
Transmission (frame or
encoder packet) error or
duplicate transmission
Explicit Rate Indication Misinterpretation of rate One or more transmission
errors due to decoding at a
different rate (modulation
and coding scheme)
User identification channel A user tries to decode a
transmission destined for
another user; a user misses
transmission destined to it.
One or more transmission
errors due to HARQ/IR
combining of wrong
transmissions
Rate or C/I feedback
channel (DRC or equivalent)
Misinterpretation of rate or
C/I for DRC feedback
information
Potential transmission errors
Fast cell site selection
signaling, e.g., transmit
sector indication, transfer of
H-ARQ states etc.
Misinterpretation of selected
sector; misinterpretation of
frames to be retransmitted.
Transmission errors
Proponents shall quantify and justify the signaling errors and their impacts in the 2
evaluation report. As an example, if an ACK is misinterpreted as a NACK (duplicate 3
transmission), the packet call throughput will be scaled down by (1-pACK), where pACK is the 4
ACK error probability. 5
2.1.6 Fairness Criteria 6
2.1.6.1 Fairness Criterion with the Normalized CDF of the User Throughput 7
Because maximum system capacity may be obtained by providing low throughput to some 8
users, it is important that all mobile stations be provided with a minimal level of 9
throughput. This is called fairness. The fairness is evaluated by determining the 10
normalized cumulative distribution function (CDF) of the user throughput, which meets a 11
predetermined function in two tests (seven test conditions). The same scheduling 12
algorithm shall be used for all simulation runs. That is, the scheduling algorithm is 13
not to be optimized for runs with different traffic mixes. The proponent(s) of any 14
proposal are also to specify the scheduling algorithm. 15
Let Tput[k] be the throughput for user k. The normalized throughput with respect to the 16
average user throughput for user k, ]k[T~
put is given by 17
3GPP2 C.R1002-A v1.0
13
][iT avg
][kT]k[T
~
puti
put
put 2.1-4 1
The CDF of the normalized throughputs with respect to the average user throughput for all 2
users is determined. This CDF shall lie to the right of the curve given by the three points in 3
Table 2.1.6-1. 4
Table 2.1.6-1 Criterion CDF 5
Normalized
Throughput w.r.t
average user
throughput
CDF
0.1 0.1
0.2 0.2
0.5 0.5
6
This CDF shall be met for the seven test conditions given in the following two tests: 7
Test 1 – for FTP, six test conditions 8
Single path Rayleigh fading 9
3, 30, 100 km/h 10
All FTP users, with buffers always full – Note that this model differs from the 11
FTP traffic model specified in section 4.1.4 12
10, 20 users dropped uniformly in a sector 13
80% (for cdma2000 1x and 1xEV-DV systems) or 100% (for 1xEV-DO 14
systems) of BS power available for data users; max. BS power = 20 w 15
Full BS power from other cells 16
The 6 test conditions are the combinations 3, 30, and 100 km/h with 10 and 17
20 FTP users per sector 18
Test 2 – for HTTP, one test condition 19
Single path Rayleigh fading 20
3 km/h 21
HTTP users, with traffic model provided in Table 2.1.6-2 – Note that this 22
traffic model differs from the HTTP traffic model specified in section 4.1.3 23
44 users dropped uniformly in a sector 24
3GPP2 C.R1002-A v1.0
14
70% (for cdma2000 1x and 1xEV-DV systems) or 100% (for 1xEV-DO 1
systems) of BS power available for data users; max. BS power = 20 w 2
Full BS power from other cells 3
4
Table 2.1.6-2 Web Browsing Model Parameters 5
Process Random Variable Parameters
Packet Calls Size Pareto with cutoff α=1.2, k=4.5 Kbytes, m=2
Mbytes, μ = 25 Kbytes
Time Between Packet
Calls
Geometric μ = 5 seconds
2.1.6.1.1 A Generic Proportional Fair Scheduler 6
Although the proponent of a proposal is free to use any scheduler, a generic proportional-7
fair scheduler [31,32], for full-buffer traffic model, with a priority function Pi(k) is given 8
below for reference: 9
( )ii
i
R kP k
T k
2.1-5 10
where k is the slot index, )(kRi is the data rate potentially achievable for the i-th mobile 11
station based upon the reported C/I and the power available to the F-PDCH, kTi is the 12
average ―fairness throughput‖ of the i-th mobile station up to time k, and is the fairness 13
exponent factor with the default value chosen as 0.75. Users with the highest priority are 14
selected for service. The number of users selected is dependent upon the number of users 15
to be serviced simultaneously. The average ―fairness throughput‖ can be calculated as 16
follows: 17
( 1) - 1
( 1) (1 ) ( 1) - 1
i
i
i i
T k if the i th MS was not scheduled at time kT k
T k N k if the i th MS was scheduled at time k
2.1-6 18
where 19
-3
3
1.25 101 for 1x EV-DV Systems
= 1.67 10
1 for 1xEV-DO Systems
t
t
2.1-7 20
t is set to 1.5s, Ni(k-1) is the number of bits delivered to the MS at time k-1 and Ti should be 21
initialized to a small value greater than zero. 22
3GPP2 C.R1002-A v1.0
15
2.1.6.2 Fairness Criterion with Geometric Mean and Harmonic Mean 1
The fundamental problem of using the fairness criterion with the normalized cdf of the user 2
throughput is that this fairness criterion takes the form of an absolute bound on relative 3
throughput. This means that it is possible that when there is extra capacity available for 4
some users but not for others, the potential extra service is disallowed, as it would violate 5
the fairness criteria. Yet it is surely always of positive benefit to be able to improve the 6
service of some users, while not reducing service on any other users. 7
A second problem is that the fairness criterion with the normalized cdf of the user 8
throughput is divorced from the final metric, the sum-throughput. The fairness criterion 9
effectively defines a set of feasible allocations, and the goal becomes to maximize the sum-10
throughput over this feasible set. This problem is generally non-trivial, and often leads to 11
iterative simulation runs searching for this optimal point for each set of simulation 12
parameters, especially since it is difficult a priori to know if a given scheduling policy will 13
lead to a feasible allocation. The complex tradeoffs made to find this ―optimal‖ allocation are 14
non-trivial and often non-transparent, and in the end are not really the appropriate 15
tradeoffs in realistic networks, due to the artificial nature of the fairness bound. 16
In this criterion a single metric is determined from the set of full-buffer throughputs which 17
applies different weighting (in the form of utility) to different levels of throughput. This 18
metric is to be optimized for each simulation scenario and used for comparison across 19
proposals. Due to the nature of the variable weighting across rates, a limitation on the 20
distribution of MS throughputs is a direct result of optimizing the metric. Hence no further 21
criterion on throughput fairness is required. This method is to be used both for pure full-22
buffer simulations and for mixed simulations which include full-buffer users. In the latter 23
case, only the throughputs of the full-buffer users are used for these comparisons. 24
There are two separate metrics that are to be used for full-buffer performance comparison, 25
the first using geometric mean (GM) and the second using harmonic mean (HM). We refer 26
to these as GMM (geometric mean method) and HMM (harmonic mean method), 27
respectively. The comparison methods under GMM and HMM are identical, other than the 28
specific metric computation. When using GMM, network resources should be allocated to 29
optimize the GM, and when using HMM they should be allocated to optimize the HM. 30
Hence, a separate simulation run is performed for the GMM and HMM comparisons. 31
Let r be the vector of throughputs from the simulation run of the full-buffer MS‘s in the 32
network, let S be the set of sectors, and let sM be the set of full-buffer MS‘s for which 33
sector s is the serving sector, and which are not in outage. The GMM metric is computed 34
as 35
Ss
N
Mi
is
t
s
s
rNN
rsec
GMM
1:)(U 2.1-8 36
where tN sec is the number of sectors in the network (57 in the standard layout), and sN is 37
the number of mobiles in sM . The HMM metric is computed as 38
3GPP2 C.R1002-A v1.0
16
Ss
Mi is
s
t
srN
NN
r11
11:)(U
sec
HMM 2.1-9 1
The following steps are to be performed for the full-buffer throughput comparison: 2
1) Determine if GMM or HMM will be used 3
2) Run the simulation (FL or RL, as desired), which includes some full-buffer users, 4
optimizing resource allocation as appropriate for GMM or HMM 5
3) From the full-buffer throughput r vector determine )(UGMM r or )(UHMM r using 6
the appropriate equation above 7
4) Report the )(UGMM r or )(UHMM r value as appropriate for comparison across 8
proposals 9
5) Report MS Relative throughput and average sector throughput as well 10
11
2.1.7 C/I Predictor Model for System Simulation 12
Each company shall use their own prediction methodology and describe the prediction 13
method in enough detail so other companies can replicate the simulations. This shall 14
include the timing diagram from measurements at the mobile to scheduling decisions at the 15
base station based on those measurements. Furthermore, this delay shall be explicitly 16
modeled in the system level simulator. 17
2.2 Link Level Modeling 18
2.2.1 Link to System FER mapping 19
The performance characteristics of individual links used in the system simulation are 20
generated a priori from link level simulations. Link level simulation parameters are 21
specified in Appendix J. 22
Turbo Decoder Metric and Soft Value Generation into Turbo Decoder shall be as specified 23
in Appendix H. 24
The quasi-static approach with fudge factors or with short term FER shall be used to 25
generate the frame erasures for both the 1xEV-DV packet data channel and the 1xEV-DO 26
Forward Traffic Channel (FTC), dynamically simulated forward link overhead channels, 27
voice and SCH (applicable only to 1xEV-DV), as described below. Equivalent SNR approach 28
shall be used to generate the frame erasures for the OFDM-based forward traffic channel 29
(FTC) in 1xEV-DO BCMCS, as described below. 30
If the BCMCS proposal uses a Reed Solomon (RS) outer code, then the frame erasures 31
generated above constitute the erasure events at the input to the outer code. This is used to 32
generate packet erasures at the output of the RS decoder as follows: 33
3GPP2 C.R1002-A v1.0
17
Suppose the RS code is specified by the ordered triple (N, K, R), where N is number of octets 1
in a RS code word, K is the number of data octets in a RS code word, and R=N-K is the 2
number of parity octets in a RS code word. Each code word of the RS code contains a data 3
octet from N consecutive turbo coded packets. A set of N consecutive turbo coded packets 4
(at the input to the RS decoder) spanned by a RS code word is referred to as an error 5
control block. An error control block contains K data packets and R parity packets. If an 6
error control block contains at most R packet erasures, then the RS outer code recovers all 7
erasures in the error control block, and all K data packets in the error control block are 8
successfully passed on to the higher layer. If the error control block contains more than R 9
packet erasures, then the unerased data packets in the error control block are passed on to 10
the higher layer, while the remaining data packets constitute erasure events at the output 11
of the RS outer code. 12
13
2.2.1.1 Quasi-Static Approach with Fudge Factors: 14
Quasi-static approach with fudge factors shall be used for 1xEV-DV Packet Data Channel, 15
1xEV-DO FTC, and Dynamically Simulated Forward Link Overhead Channel. 16
The aggregated Es/Nt is computed over a transmission period and mapped to an FER using 17
AWGN curves. The proponent shall select one of two possible methods to determine the 18
FER: 19
a) Map the aggregated Es/Nt directly to the AWGN curve corresponding to the given 20
modulation and coding. 21
b) Adjust the aggregated Es/Nt for the given modulation and coding and lookup a 22
curve obtained using a reference modulation and coding. 23
Furthermore the proponents shall account for an additional Es/Nt loss at higher Dopplers 24
for either method. 25
Full details of the quasi-static frame error modeling with fudge factors are given in the 26
Appendix F. 27
2.2.1.2 Quasi-Static Approach with Short Term FER: 28
The quasi-static approach with short term FER may be used to generate frame erasures for 29
voice, SCH, and F-PDCH for 1xEV-DV systems. The quasi-static approach with short term 30
FER may be used to generate frame (i.e. physical-layer packet) erasures for the Forward 31
Traffic Channel (FTC) for 1xEV-DO systems. 32
A full set of short term FER vs. average Eb/Nt per frame curves is generated as a function 33
of radio configurations, transmission diversity schemes (if applicable), channel models, 34
different ways of soft hand-off (SHO), different SHO imbalances, and geometries. The 35
number of curves should be reduced if possible, provided that this won‘t unduly affect the 36
validity of this quasi-static approach. 37
All companies shall use the same set of short term FER vs. average Eb/Nt per frame. 38
3GPP2 C.R1002-A v1.0
18
In the system-level simulation, the average Eb/Nt per frame is computed as follows. First, 1
the average Eb/Nt is calculated in a PCG (slot). The short-term average Eb/Nt per frame is 2
defined as the average of the average Eb/Nt for all Ns PCG‘s (slots) in a frame (physical 3
layer packet), i.e., 4
sN
n nt
b
st
b
N
E
NN
E
1
1 2.2-1 5
where (Eb/Nt)n is the average Eb/Nt in the n-th PCG (slot) in a frame (physical-layer 6
packet). Note that finger combining and self-interference are applied before the average 7
Eb/Nt in a PCG (slot) is calculated. Once the Eb/Nt is calculated as in the above equation, 8
it is used to look up the corresponding link level short term FER vs. average Eb/Nt per 9
frame curves for the specific condition (i.e., radio configuration, transmission diversity (if 10
applicable) scheme, channel model, way of soft hand-off (SHO), SHO imbalance(s), and 11
geometry). A frame erasure event is then generated based on the FER value. 12
If a short term FER vs. average Eb/Nt per frame curve is not available for a condition, the 13
curve should be computed by interpolating those curves for similar conditions (e.g., 14
between the factors for closest geometries available). 15
The short term FER vs. average Eb/Nt per frame curves shall be generated as follows: 16
1. The link-level simulation is conducted for a specific condition. The average Eb/Nt in a 17
frame and the frame erasure indicator for the frame are recorded. For 1xEV-DV 18
systems, the average Eb/Nt per frame is computed as follows in the link-level 19
simulation 20
sN
n
k
kn
t
k
kn
b
t
b
n
Sm
N
E
12),(
2),(
16
1 2.2-2 21
where n is the index of PCG in a frame and k is the index of symbols within a PCG. 22
),( kn
bS is the signal component in the k-th received coded symbol in the n-th PCG, ),( kn
tn 23
is the noise and interference component in the k-th received symbol in the n-th PCG in 24
a frame, and m is the inverse of the code rate (i.e., 4 for RC3 and 2 for RC4, etc). For 25
1xEV-DO systems, the average Eb/Nt per slot is computed as follows in the link-level 26
simulation 27
K
k
L
llkn
t
lkn
sn
t
b
N
E
MN
E
1 1),,(
),,(1
)( 2.2-3 28
where M equals to the number of information bits per packet; n is the slot index and 29
k is the symbol index within a slot; l is the path index, where the total number of 30
captured paths is denoted by L; the total number of symbols per slot is K; 31
),,( lkn
sE denotes the signal energy in the k-th received symbol in the n-th slot in the 32
physical-layer packet at the l-th RAKE finger, and ),,( lkn
tN denotes the noise and 33
interference variance in the k-th received symbol in the n-th slot in the physical-34
layer packet seen at the l-th RAKE finger. 35
3GPP2 C.R1002-A v1.0
19
2. Generate the histogram of FER vs. the average Eb/Nt per frame, i.e., the range of Eb/Nt 1
is divided into many bins, and the FER in each bin is computed based on the outputs 2
mentioned in step 1. The size of each bin is 0.25 dB. 3
4
2.2.1.3 Equivalent SNR Approach: 5
Equivalent SNR approach defined in Appendix T shall be used to generate the frame 6
erasures for the OFDM-based forward traffic channel (FTC) in 1xEV-DO BCMCS. 7
The equivalent SNR of the frame is computed over a transmission period and mapped to an 8
FER using the AWGN reference curve. The AWGN reference curves are obtained by 9
simulating the turbo code performance on an AWGN channel for each transmission format, 10
defined by payload size and number of slots. The equivalent SNR of each frame is 11
calculated from the SNR per tone at the output of the FFT demodulator. The SNR per tone 12
is calculated considering coherence loss due to Doppler, inter-tone interference, and 13
channel estimation loss. 14
Equivalent SNR Method based on Convex Metric (ECM) described in Appendix O shall be 15
used to generate frame erasures for UMB system. Justification on the parameter values 16
used shall be provided by each proponent. The link level statistics that are used for 17
generating the short-term FER curves for link-to-system mapping is shown in Appendix W. 18
19
2.2.2 Channel Models 20
2.2.2.1 Channels model based on ITU channel model 21
A channel model corresponds to a specific number of paths, path delay and power profile 22
(ITU multi-path models), and Doppler frequencies for the paths. 23
Table 2.2.2-1 Channel Models 24
Channel
Model
Multi-path
Model
# of Fingers Speed
(kmph)
Fading Assignment
Probability
Model A Pedestrian A 1 3 Jakes 0.30
Model B Pedestrian B 3 10 Jakes 0.30
Model C Vehicular A 2 30 Jakes 0.20
Model D Pedestrian A 1 120 Jakes 0.10
Model E Single path 1 0, fD=1.5 Hz Rician Factor
K = 10 dB
0.10
25
The channel models are randomly assigned to the various users according to the 26
probabilities of Table 2.2.2-1 at the beginning of each drop and are not changed for the 27
duration of that drop. The assignment probabilities given in Table 2.2.2-1 are interpreted 28
3GPP2 C.R1002-A v1.0
20
as the percentage of users with that channel model in each sector. The JTC fader (see 1
Appendix K) shall be used to generate the Jakes fading samples. 2
For CDM transmissions with Rake demodulation, each multipath model (Pedestrian A, 3
Vehicular A/B etc) is characterized in terms of the number of Rake fingers (resolvable 4
paths), the Delay and Fractional Recovered Power (FRP) of each finger, and the Fractional 5
UnRecovered Power (FURP). The FRP and FURP are given in Table 2.2.2-2. FURP shall 6
contribute to the interference of the finger demodulator outputs as an independent fader. 7
The power on all fingers (including FURP) for each channel model shall be normalized so 8
that the total power for that channel model adds up to unit one. 9
Table 2.2.2-2 Fractional Recovered Power and Fractional UnRecovered Power 10
Model Finger1
(dB)
Delay Finger2
(dB)
Delay (Tc) Finger3
(dB)
Delay (Tc) FURP (dB)
Ped-A -0.06 0.0 -18.8606
Ped-B -1.64 0.0 -7.8 1.23 -11.7 2.83 -10.9151
Veh-A -0.9 0.0 -10.3 1.23 -10.2759
11
The delay values given in Table 2.2.2-2 are for information purposes and do not need to be 12
accounted for in the system simulation. 13
For OFDM transmission and demodulation of 1xEV-DO BCMCS, each multipath model 14
(Pedestrian A, Vehicular A/B etc) is characterized in terms of the total number of paths, 15
together with actual power-delay profile of the multipath channel. For each multipath 16
model, the power on all paths shall be normalized so that the total power adds up to one. 17
The parameters of the multipath models for OFDM transmission and demodulation are 18
given in Table 2.2.2-3 and Table 2.2.2-4. 19
Table 2.2.2-3 Relative Power of each Multipath Model (in dB) 20
Model # Paths 1 2 3 4 5 6
Ped-A 4 0 -9.7 -19.2 -22.8
Ped B 6 0 -0.9 -4.9 -8.0 -7.8 -23.9
Veh-A 6 0 -1.0 -9.0 -10.0 -15.0 -20.0
21
Table 2.2.2-4 Delay of each Multipath Model (in ns) 22
Model # Paths 1 2 3 4 5 6
Ped-A 4 0 110 190 410
Ped B 6 0 200 800 1200 2300 3700
Veh-A 6 0 310 710 1090 1730 2510
3GPP2 C.R1002-A v1.0
21
The channel between the serving sector(s) and the subscriber is modeled using the 1
multipath profiles defined above. The channel between any interfering sector and the 2
subscriber can be modeled as a one-path Rayleigh fading channel, where the Doppler of the 3
fading process is randomly chosen based on the velocities and its corresponding 4
probabilities specified in Table 2.2.2-1. 5
2.2.2.2 Channels model based on SCM 6
2.2.2.2.1 Channel model for system level simulations 7
The 3GPP2/3GPP spatial channel model (SCM) [41] shall be used for all system level 8
simulations. The Urban Macro-cellular environment is mandatory and the parameters of 9
Table 2.2.2-5 shall be used for configuring the model. 10
Table 2.2.2-5 Macro-cellular Environment Parameters 11
Channel Scenario Urban Macro
Number of paths (N) 6
Number of sub-paths (M) per-path 20
Mean AS at BS E( AS )=15 deg
AS at BS as a lognormal RV
10 ^ , ~ (0,1)AS AS ASx x
15deg
AS = 1.18, AS = 0.210
ASAoDASr / 1.3
Per-path AS at BS (Fixed) 2 deg
BS per-path AoD Distribution standard
distribution
),0( 2
AoD where ASASAoD r
Mean AS at MS E(AS, MS)=68 deg
Per-path AS at MS (fixed) 35 deg
MS Per-path AoA Distribution (Pr)),0( 2
AoA
Delay spread as a lognormal RV
10 ^ , ~ (0,1)DS DS DSx x
DS = -6.18
DS = 0.18
Mean total RMS Delay Spread E( DS )=0.65 s
DSdelaysDSr / 1.7
Lognormal shadowing standard deviation,
SF
8.9dB
Pathloss model (dB), d is in meters 28.6 + 35log10(d)
3GPP2 C.R1002-A v1.0
22
The velocity profile is shown in Table 2.2.2-6. Because of the choice of urban macrocell, 1
velocities are biased towards pedestrian speeds. 2
3
Table 2.2.2-6 Quantized Velocity Profile 4
Percentage Velocity (km/h)
35% 3
30% 30
20% 60
15% 120
The carrier frequency for all simulations is assumed to be 2.0 GHz. 5
2.2.2.2.2 Channel model for link level simulations 6
For link-level simulations, that excludes link to system mapping simulations, spatially 7
extended ITU profiles will be used. 8
Table 2.2.2-7 ITU Profiles for Link Level Simulations 9
ITU Model Velocity (km/h)
AWGN 0
Ped-A 3
Ped-B {3,30}
Veh-A {30,120}
For technologies that use a cyclic prefix, the path location will be equal to the closest 10
integer sample number, i.e. no FURP modeling is needed. 11
12
Table 2.2.2-8 ITU Profiles Spatial Extension Parameters 13
Channel Scenario Urban Macro
AS at BS AS =150
Per-path AS at BS (Fixed) 2 deg
AS at MS AS, MS=680
Per-path AS at MS (fixed) 350
AoDs As specified in Table 2.2.2-9
AoAs As specified in Table 2.2.2-9
14
3GPP2 C.R1002-A v1.0
23
Table 2.2.2-9 Path power, AoD, AoA 1
Path Power Path AOD (rad) Path AoA (rad)
Ped-A 0.889345301 0.346314033 1.737577272
0.095295066 -0.05257642 -1.55645
0.010692282 -1.817837659 -1.049078459
0.00466735 -0.836999548 0.345571431
Ped-B 0.405688403 -0.13638548 1.319340881
0.329755914 0.302249557 -0.119072067
0.131278194 0.496051618 0.901442565
0.064297279 0.544719913 -1.424448314
0.067327516 0.212670549 -3.062670939
0.001652695 -0.604134536 -1.202289294
Veh-A 0.48500285 -0.46084874 -0.780118399
0.385251458 -0.897480352 -1.729577654
0.061058241 -0.525726742 1.792547973
0.048500285 0.00282531 1.776985779
0.015337137 -1.016095677 1.386034573
0.004850029 0.245512493 3.50389557
The fading coefficient generation will be the one described in section 5.3 of the 2
3GPP2/3GPP SCM [41]. 3
2.2.2.2.3 Channel model for virtual decoder generation and verification 4
The 3GPP2/3GPP2 SCM channel model for the urban macro environment will be used for 5
generating the Virtual Decoder and Fudge Factors. The velocities of 3, 30 and 120 km/h 6
will be used. The values in Appendix V represent 64 realizations of the SCM urban macro 7
environment. 8
2.2.3 C/I modeling for system simulation 9
Each channel model shall be modeled in the system level simulation as follows: 10
In the system level simulation for CDM transmissions with Rake demodulation, the 11
interference due to unrecovered power shall be modeled as an additional ray that is not 12
demodulated by the Rake receiver. Let J denote the number of rays used in a particular 13
channel model, excluding the ray used to model FURP. The average power assigned to each 14
of the rays is given in Table 2.2.2-2. The average power assigned to the ray used to model 15
FURP is also given in Table 2.2.2-2. The recovered rays and the additional ray used to 16
model the unrecovered power all fade independently of each other. 17
3GPP2 C.R1002-A v1.0
24
Let J
ii 1 denote the samples of the fading processes, for a particular PCG, of the J 1
recovered rays. Let denote the sample of the fading process for the additional ray used to 2
model interference due to the unrecovered power, for a particular PCG. Let J
iiIC
1 3
denote the signal-to-interference ratio for each of the Rake fingers, which can be expressed 4
as 5
ikJk
k
i
iG
IC
,1
221
2
2.2-4 6
where G denotes the subscriber geometry, given by 7
N
n
noc
or
nIN
IG
1
2
0 )(
ˆ
2.2-5 8
N is the number of interfering sectors, n is the fading process of the ray between the 9
receiver and the n-th interfering sector for a particular PCG, N0 is the variance of the 10
thermal noise including the mobile station noise figure defined in Table 2.1.2-1, orI is the 11
total energy per chip averaged over fading and received from the serving sector, and Ioc(n) is 12
the total energy per chip averaged over fading and received from the n-th interfering sector. 13
In the system level simulation, the Rake fingers shall be combined using pilot-weighted 14
combining. The signal-to-interference ratio at the output of the pilot-weighted combiner is 15
given by 16
J
j jkJk
kj
J
i
i
G
IC
1 ,1
2212
2
1
2
combined
2.2-6 17
This combined C/I shall further be limited by a C/I ceiling as described in section 2.1.2. 18
For system level simulations including transmit diversity (STS), the channels between the 19
two transmit antennas and the subscriber are assumed to fade independently of each 20
other. The channel models are taken from Table 2.2.2-1. For a particular PCG, let J
ii 1 21
and J
ii 1
~
, respectively, denote the samples of the fading processes for the J recovered rays 22
of the first and second antennas. Let and ~
, respectively, denote the sample of the 23
fading process for the additional rays used to model interference due to the unrecovered 24
power for the first and second antennas. 25
Let J
iiIC
1,1 denote the signal-to-interference ratio of Rake fingers demodulating symbols 26
transmitted from the first antenna. For transmit diversity using STS, one-half of the energy 27
of a given code symbol is transmitted on each of the antennas. If all code channels are 28
3GPP2 C.R1002-A v1.0
25
transmitted using transmit diversity, the signal-to-interference ratio of the i-th Rake finger 1
can be expressed as 2
ikJk
kk
i
i
G
IC
,1
22221
2
,1
~~
2
1
2
2.2-7 3
Let J
iiIC
1,2 denote the signal-to-interference ratio of Rake fingers demodulating symbols 4
transmitted from the second antenna, which can be expressed as 5
ikJk
kk
i
i
G
IC
,1
22221
2
,2
~~
2
1
2~
2.2-8 6
The signal-to-interference ratio for STS with pilot-weighted combining of the Rake fingers in 7
both delay and diversity is given by 8
J
j jkJk
kkjj
J
i
ii
G
IC
1 ,1
2222122
2
1
22
combinedSTS,
~~
2
1~
~
2
1
2.2-9 9
This combined C/I shall further be limited by a C/I ceiling as described in section 2.1.2. 10
11
The alternative approach to obtain the combined C/I in the system level simulation for 12
CDM transmissions with Rake demodulation is shown in Appendix S. This approach can be 13
used for 1xEV-DO BCMCS. 14
15
In the simulation for OFDM transmission and demodulation of 1xEV-DO BCMCS, the SNR 16
of each tone at the output of the FFT operation in the OFDM demodulator is computed as 17
follows. This model assumes all cells are in broadcast mode, transmitting the same OFDM 18
symbols. 19
Let Ls, As and Ss denote the path loss, antenna gain and log-normal shadowing process, 20
respectively, from the s-th sector to the mobile station. Let js denote the absolute delay of 21
j-th path from the s-th sector and let ijs denote the complex channel gain of the fading 22
process for this path to the i-th receiver antenna (2-way diversity is assumed with i=[0,1]). 23
The absolute delay js may be expressed as js
jsc
d , where sd denotes the distance of 24
the s-th sector from the mobile station, c denotes the speed of light, and j denotes the 25
delay of the j-th multipath component, as specified in Table 2.2.2-4. The complex channel 26
3GPP2 C.R1002-A v1.0
26
gain ijs is modeled as a (Jakes or Ricean) fading process with Doppler specified in Table 1
2.2.2-1 and power of the j-th multipath component as specified in Table 2.2.2-3. 2
Let Ci(f) denote the baseband frequency response of composite channel from all sectors s to 3
the i-th diversity antenna. The composite channel can be expressed in terms of ijs and js 4
as follows 5
jsfi
ijssssi eSALfC
2
)( 2.2-10 6
Let )( fG denote the frequency response of the baseband transmit pulse shaping filter and 7
assume the receiver filter is matched to )( fG . 8
Let )(kI , k=[-N/2,…,N/2-1], denote the complex information symbol transmitted at tone k 9
at baseband frequency of NT
kf k , where N is the OFDM symbol length and T is the chip 10
period. Default parameters are 320N and 6102288.1/1 T . 11
The effective channel response Hi,k associated with the k-th OFDM tone at the i-th diversity 12
antenna is given by 13
T
mfC
T
mfG
TH ki
m
kki
2
,
1 2.2-11 14
Note that 1,0,1m is sufficient to capture the significant alias components since )( fG is 15
approximately band-limited to [-1/2T, 1/2T]. 16
Following cyclic prefix extraction and FFT operation at the demodulator, the corresponding 17
received sample ][kYi for tone k is given by: 18
][][][ , kNkIEHkY ckkii 2.2-12
19
where cE is the transmit energy per chip, k is the normalized tone power, Hi,k is the 20
effective channel response of the k-th tone at the i-th diversity antenna, and )(kN is a 21
complex Gaussian noise term corresponding to the receiver noise floor(including thermal 22
noise floor and SNR ceiling of the receiver). The normalized tone power k is the ratio of the 23
power allocated to the k-th tone to the average power of the tones. Typically, k will be 24
greater than 1.0 due to the presence of guard tones. For a system with 320 subcarriers and 25
16 guard tones, if the signal energy is allocated uniformly among the 304 non-guard tones, 26
then the normalized tone power of any non-guard tone is given by 304
320k . The variance 27
of )(kN is determined by the autocorrelation of the receiver filter as follows 28
22][ koNkNE
2.2-13
29
3GPP2 C.R1002-A v1.0
27
max
1
0 0
2
,2
2
)0(1
1
I
C
RN
EH
NejTR
N
jg
N
m
cmmi
N
jki
g
N
Nj
k
2.2-14
1
where )(tRg is the autocorrelation of the baseband pulse g(t) 2
dssgstgtRg )()()( *
2.2-15
3
and (C/I)max denotes the maximum achievable C/I for the subscriber receiver, as specified 4
in Table 2.1.2-1 and Table 2.1.2-4 5
The demodulated symbol SNR of the k-th tone for the i-th diversity branch is given by 6
WN
PH
N
EHkiSNR TXk
k
kick
k
ki
0
2
2
,
0
2
2
,],[
2.2-16
7
where PTX is the total per sector transmit power, W is the chip-rate, and N0 is the thermal 8
noise spectral density at the subscriber receiver. 9
10
2.3 Simulation Flow and Output Matrices 11
Either the center cell method or the iteration method shall be used. 12
The total simulation time per drop shall be long enough to guarantee residual FER to be 13
10-2 with certain confidence level. The required upper layers to guarantee the TCP input 14
FER to be 10-4 is not modeled. 15
2.3.1 Simulation Flow for the Center Cell Method 16
The simulation will make the following assumptions: 17
1. The system consists of 19 hexagonal cells. Six cells of the first tier and 12 cells of 18
the second tier surround the central cell. Each cell has three sectors. 19
2. Mobiles are first dropped uniformly throughout the system. Each mobile 20
corresponds to an active user session. A session runs for the duration of the drop. 21
Mobiles are assigned channel models described in Table 2.2.2-1 with the given 22
probabilities. 23
3. Applicable to 1xEV-DV only: For simulation of systems loaded only with voice-only 24
mobiles, the first run is done with five mobile stations per sector and every run 25
following that is done with five more mobile stations per sector until the outage 26
criteria are violated, after which every run is done with one less mobile until the 27
outage criteria are satisfied. The maximum number of users per sector of Nmax (voice 28
capacity) is achieved if Nmax + 1 users per sector would not satisfy the outage criteria 29
and Nmax users per sector would. 30
3GPP2 C.R1002-A v1.0
28
4. For simulation of the system loaded only with data-only mobiles, the runs are done 1
with an increment of two mobile stations per sector. 2
5. Applicable to 1xEV-DV only: With voice-only and data-only mobile stations in the 3
same simulated system, the number of voice-only mobile stations is fixed at 4
0.5Nmax or 0.8Nmax per sector. The number of data-only mobile stations is 5
incremented by two mobile stations per sector for each successive simulation run. 6
6. Mobile stations are randomly dropped over the 57 sectors such that each sector has 7
the required numbers of voice (1xEV-DV only) and data users. Although users may 8
be in soft-handoff, each user is assigned to only one sector for counting purposes. A 9
data user shall be assigned to a sector if the sector is its primary server. To simplify 10
the 1xEV-DV simulation, only two-way or three-way handoff is used for voice users. 11
A 1xEV-DV voice user on a two-way or three-way soft handoff counts as ½ or 1/3 a 12
user on each of the sectors in the Active Set, respectively. All sectors of the system 13
shall continue accepting users until the desired fixed number of data/voice users 14
per sector is achieved everywhere. 15
7. Fading signal and fading interference are computed from each data/voice mobile 16
station into each sector for each PCG (slot) or equivalent power control related time 17
interval. 18
8. The total simulation time per drop will be 10 minutes excluding any time required 19
for initialization (~10 sec). The total number of drops per run is 12 for a total 20
simulation time of 2 hours per condition. 21
9. Packet calls arrive as per the HTTP model of section 4.1.3. Packets are not blocked 22
when they arrive into the system (i.e. queue depths are infinite). 23
10. WAP is modeled as per section 4.1.5. 24
11. FTP is modeled as per section 4.1.4. FTP results presented shall be from a stable 25
system. That is a system in which the average rate of FTP users exiting the system 26
is equal to the average rate of FTP users entering the system. 27
12. Near real time video is modeled as per section 4.1.6. 28
13. Packets are scheduled with a packet scheduler. 29
14. The ARQ process is modeled by explicitly rescheduling a packet as part of the 30
current packet call after a specified ARQ feedback delay period. 31
15. Simulation flow with data-only mobiles or voice-only and data-only mobiles 32
simultaneously is shown in Figure 2.3.1-1: 33
For n = 0, 0.5Nmax or 0.8Nmax voice-only mobiles per sector (1xEV-DV only), 34
For each k data-only users (to be incremented by 2), 35
a. Place n voice users in each sector of all cells (1xEV-DV only). 36
b. Keep adding data users until the quality of service criteria are not met. 37
c. Collect results according to the output matrix. 38
3GPP2 C.R1002-A v1.0
29
16. Only statistics of the mobiles of the center cell are collected. 1
17. All 57 sectors in the system shall be dynamically simulated. 2
Compute Average
Service Throughput
Load System with k Data
Mobile per Sector
Fairness & Outage
Criteria Satisfied?
Increment k by 2
Yes
STOP
No
Load System with 0,
0.5Nmax or 0.8Nmax
voice Mobiles per Sector } Voice applicable to
1xEV-DV
systems only
3
Figure 2.3.1-1 Simulation Flow Chart 4
2.3.2 Simulation Flow for the Iteration Method 5
The cells are classified into two types: active and passive cells. Active cells are fully 6
simulated and monitored, whereas passive cells only model the interference created by 7
neighboring cells. 8
3GPP2 C.R1002-A v1.0
30
Terminology: 1
Active cell – the central cell of the 19 cell layout which handles all the data traffic, runs a 2
scheduler, keeps track of the voice users (when applicable), generates the 3
transmission power profile, and collects all the statistics. 4
Passive cells – the other 18 cells, which follow the power profile of the active cell as found 5
in the previous iteration. In order to break the temporal correlations 18 equally 6
spaced offsets are introduced, one for each passive cell. For example, for 600 7
second run each offset is 33 seconds, which is sufficient to compensate for any 8
correlation. For 1xEV-DV systems, voice calls can be in soft handoff with passive 9
cells, in which case Ec/Ior for a given call is assumed to be the same at all the base 10
stations in the active set. Passive cells are present in the system in order to model 11
the inter-cell interference, and therefore data users associated with them do not 12
need to be modeled. 13
Iteration – A simulation run where passive cells follow a well-specified power profile 14
obtained from the active cell on a previous iteration. Iteration 0 starts with passive 15
cells transmitting with the maximum power. 16
Description: 17
The system consists of 19 hexagonal cells. The central cell is an active cell and is 18
surrounded by six passive cells of the first tier and 12 passive cells of the second tier 19
cells. Each cell has three sectors. 20
Applicable to 1xEV-DV only: For simulation of systems loaded only with voice-only 21
mobiles, the first run is done with five mobile stations per sector and every run 22
following that is done with five more mobile stations per sector until the outage criteria 23
are violated, after which every run is done with one less mobile until the outage criteria 24
are satisfied. The maximum number of users per sector of Nmax (voice capacity) is 25
achieved if Nmax + 1 users per sector would not satisfy the outage criteria and Nmax users 26
per sector would. 27
For simulation of the system loaded only with data-only mobiles, the runs are done with an 28
increment of two mobile stations per sector. 29
Applicable to 1xEV-DV only: With voice-only and data-only mobile stations in the same 30
simulated system, the number of voice-only mobile stations is fixed at 0.5Nmax or 31
0.8Nmax per sector. The number of data-only mobile stations is incremented by two 32
mobile stations per sector for each successive simulation run. 33
Voice and data users are randomly dropped over the 57 sectors such that each cell has the 34
required numbers of voice and data users. To simplify the 1xEV-DV simulation, only 35
two-way or three-way handoff is used for voice users. A 1xEV-DV voice user on a two-36
way or three-way soft handoff counts as ½ or 1/3 a user on each of the sectors in the 37
Active Set, respectively. A data user shall be assigned to a sector if the sector is its 38
primary server. Users whose Active Set does not contain a sector of the active (center) 39
cell shall be discarded. The sectors of the active cell shall continue accepting users until 40
the desired fixed number of data/voice users per sector is achieved. 41
3GPP2 C.R1002-A v1.0
31
Fading signal and fading interference are computed for each data/voice user for each PCG 1
or equivalent power control related time interval. 2
Iterations are performed in the following order: 3
Iteration 0: Passive cells radiate at maximum power. Power statistics of the active 4
(central) cell is collected for use in the next iteration. 5
Iteration n (n>0): Run the system forcing passive cells to follow the active‘s cell power 6
profile found on the iteration (n-1). Time offsets are introduced to break the correlation, 7
as described previously. 8
Convergence criterion: stability of the per sector throughput and of the power profile second 9
order statistics 10
The total simulation time per drop will be 10 minutes excluding any time required for 11
initialization (~10 sec). The total number of drops per run is 12 for a total simulation 12
time of 2 hours per condition. 13
Packet calls arrive as per the HTTP model of section 4.1.3. Packets are not blocked when 14
they arrive into the system (i.e. queue depths are infinite). 15
WAP is modeled as per section 4.1.5. 16
FTP is modeled as per section 4.1.4. 17
Near real time video is modeled as per section 4.1.6. 18
Packets are scheduled with a packet scheduler. 19
The ARQ process is modeled by explicitly rescheduling a packet as part of the current 20
packet call after a specified ARQ feedback delay period. 21
Simulation flow with data-only mobiles or voice-only and data-only mobiles simultaneously 22
is shown in Figure 2.3.1-1: 23
For n = 0, 0.5Nmax or 0.8Nmax voice-only mobiles per sector (1xEV-DV only), 24
For each k data-only users (to be incremented by 2), 25
a. Place n voice users in each sector of all cells. 26
b. Keep adding data users until the quality of service criteria are not met. 27
c. Collect results according to the output matrix. 28
2.3.3 Simulation Flow for the Wrap Around Method 29
There are 19 3-sectored cells in the simulated system. All cells are fully simulated and the 30
statistics of the mobiles of all cells are collected. See Appendix I for details. 31
The simulation for 1xEV-DO BCMCS will make the following assumptions: 32
1. The system consists of 19 hexagonal cells. Six cells of the first tier and 12 cells of 33
the second tier surround the central cell. Each cell has three sectors. 34
2. Mobile stations are uniformly dropped into the 19-cell system. A mobile station 35
dropped within 35 meters of a base station is redropped. One fixed path loss and 36
3GPP2 C.R1002-A v1.0
32
one randomized shadowing component are computed for each BS-MS pair for the 1
duration of the simulation run. Two independent forward link fading components 2
are computed once every slot for the two mobile station receiving antennas for each 3
base station. 4
3. A MS dropped into the system shall be redropped if there are no sectors in its active 5
set, or if the total path loss to every sector is above a threshold defined in Table 6
2.1.2-1. 7
4. The BCMCS traffic shall be modeled as an infinite queue, that is transmitted on the 8
forward link once every sixteen slots (one interlace-multiplex pair). The same 9
BCMCS data is transmitted by all the base stations at the same time. 10
5. Fading signal and fading interference are computed from each mobile station into 11
each sector for each slot. 12
6. Since there is no interaction between mobile stations in the BCMCS system 13
simulation, mobile stations may be simulated in the system either concurrently in 14
one simulation, or individually in multiple simulation runs. 15
7. The total simulation time per mobile station determines the confidence interval in 16
the packet error rate (PER) estimated by the simulation. To ensure sufficient 17
confidence in the PER statistics, at least 100,000 packets are simulated per mobile 18
station. 19
8. The proponents shall provide the 95% confidence interval associated by the system 20
coverage (for a given data rate, at a given PER target) estimated by the simulation.. 21
Details of the confidence margin calculation (for a given number of users in the 22
simulation) may be found in [34]. 23
2.3.4 Layout Files 24
A pair of layout files is used for all the network simulations in UMB, one for AN and one for 25
AT layout. The AN layout file contains one line per sector, and specifies sector number, 26
sector location, antenna angle, and associated cell number. The AT layout file contains one 27
line per AT, and specifies AT layout number, the AT‘s best sector, AT location, path loss 28
from the AT to each of the 57 sectors, and channel type, with optional SCM parameter 29
fields. The path loss includes propagation loss, shadowing, and antenna gain for that AT 30
location. 31
The layout files represent a database which is referenced for each network simulation to 32
provide a set of AT locations. These references are made in the source configuration file for 33
each simulation run, where each location is combined with a source description for each AT. 34
The idea is that one pair of layout files is used for all runs, with locations and sources 35
described for each run in the source description file for that run. 36
The following data is in ASCII text format. Details of the format of these layout files are 37
shown below: 38
39
40
3GPP2 C.R1002-A v1.0
33
AN Layout File
Sector# Xloc Yloc AntAngle Cell#
Sector#: Sector reference index (1-57)
Xloc: Sector x-axis location (km)
Yloc: Sector y-axis location (km)
AntAngle: Sector antenna orientation (-30, 90, 210 deg)
Cell#: Index of sector‘s cell(1-19)
AT Layout File
AT Lay# BestSect Xloc Yloc S1PL S2PL … S57PL ChanType [SCMParam]
AT Lay#: Reference # used in source configuration file (starts at 1)
BestSect: Best sector # for this AT (1-57)
Xloc: x-axis location (km)
Yloc: y-axis location (km)
SnnPL: Path loss to sector nn, including propagation loss, shadowing, and antenna gain
(dB)
ChanType: 1=A, 2=B, 3=C, 4=D, 5=E
SCM: TBD
1
2.3.5 Output Matrices 2
The performance report shall contain a pdf of the forward link C/I observed in one of the 3
sectors of the center cell under the assumption of full power transmitted by all sectors. 4
Three curves shall be generated. First one includes path loss, shadowing, sectorization, 5
but not Rayleigh fading; the second one and the third one are the first one with the 6
restriction of maximum C/I to be 13.5 dB and 17.8 dB for CDM transmissions with Rake 7
demodulation, respectively. For OFDM transmissions of 1xEV-DO BCMCS, the first one 8
with the restriction of maximum C/I of 17 dB shall be generated. 9
Table 2.3.5-1 summarizes all the cases to be simulated for 1xEV-DV systems. 10
3GPP2 C.R1002-A v1.0
34
Table 2.3.5-1 Required 1xEV-DV Simulation Evaluation Comparison Cases Table 1
2
3
The voice capacity numbers generated from run #1 to run #8 should be approximately the 4
same across different companies, which can be used as calibration of different simulators. 5
1xEV-DV data only with transmit diversity scenarios (run #9 and run #11) are optional for 6
proposals that do not support transmit diversity. 7
2.3.5.1 General output matrices 8
The following matrices shall be provided: 9
1. All link-level results used in system-level simulator, 10
2. The histogram of C/I used in system-level simulation, where the C/I includes path 11
loss, shadowing, and sectorization: 12
a. Without any limitation on the maximum C/I 13
b. With a limit of 13.5 dB on the maximum C/I, as specified in section 2.1.2 14
Tx Diversity no Tx Diversity Max C/I 13.5 dB Max C/I 17.8 dB RC3 RC4 Loading Scenarios
1 voice only 100% (Nmax) load x x x
2 x x x
3 x x x
4 x x x
5 x x x
6 x x x
7 x x x
8 x x x
9 1xEVDV data only x x
10 x x
11 x x
12 x x
13 50%voice + 1xEVDV data x x x
14 x x x
15 x x x
16 x x x
17 x x x
18 x x x
19 x x x
20 x x x
21 80%voice + 1xEVDV data x x x
22 x x x
23 x x x
24 x x x
25 x x x
26 x x x
27 x x x
28 x x x
3GPP2 C.R1002-A v1.0
35
c. With a limit of 17 dB on the maximum C/I, as specified in section 2.1.2 1
(Applied only for OFDM transmission of 1xEV-DO BCMCS) 2
d. With a limit of 17.8 dB on the maximum C/I, as specified in section 2.1.2 3
3. The curve of geometry vs. the distance from a user‘s location to its closest serving 4
cell, where geometry is solely a function of the distance and excludes fading and 5
shadowing factors. 6
4. The histogram of the distance between a user and its closest serving cell/sector. 7
5. The performance of any estimator(s) or predictor(s) that are required by the 8
proposal. For instance, if a channel predictor is used in the proposal, the details of 9
the predictor/estimator, its bias, and standard deviation shall be provided by the 10
proponent(s). 11
6. The performance of any forward channels that are not simulated by the link-level or 12
system-level simulator shall be justified by the proponent. For example, if the 13
forward signaling channels are not simulated by the system-level simulator, the 14
proponent companies shall specify the performance of these channels and justify 15
their claims (see section 2.1.5). 16
2.3.5.2 Data Services and Related Output Matrices 17
The following statistics related to data traffics shall be generated and included in the 18
evaluation report. 19
1. Data throughput per sector. The data throughput of a sector is defined as the 20
number of information bits per second that a sector can deliver and are received 21
successfully by all data users it serves, using the scheduling algorithm validated in 22
section 2.1.6, and that certain number of voice users can be maintained with 23
certain GOS. 24
2. Averaged packet delay per sector. The averaged packet delay per sector is defined 25
as the ratio of the accumulated delay for all packets it delivers to all users and the 26
total number of packets it delivers. The delay for an individual packet is defined as 27
the time between when the packet enters the queue at transmitter and the time 28
when the packet is received successively by the mobile station. If a packet is not 29
successfully delivered by the end of a run, its ending time is the end of the run. 30
3. The histogram of data throughput per user. The throughput of a user is defined 31
as the ratio of the number of information bits that the user successfully receives 32
during a simulation run and the simulation time. Note that this definition is 33
applicable to all data users. 34
4. The histogram of packet call throughput for users with packet call arrival 35
process. The packet call throughput of a user is defined as the ratio of the total 36
number of information bits that an user successfully receives and the accumulated 37
delay for all packet calls for the user, where the delay for an individual packet call is 38
defined as the time between when the first packet of the packet call enters the 39
queue for transmission at transmitter and the time when the last packet of the 40
3GPP2 C.R1002-A v1.0
36
packet call is successively received by the receiver. If a packet call is not 1
successfully delivered by the end of a run, its ending time is the end of the run, and 2
none of the information bits of the packet call shall be counted. Note that this 3
definition is applicable only to a user with packet call arrival process. 4
5. The histogram of averaged packet delay per user. The averaged packet delay is 5
defined as the ratio of the accumulated delay for all packets for the user and the 6
total number of packets for the user. The delay for a packet is defined as in 2. Note 7
that this definition is applicable to all data users. 8
6. The histogram of averaged packet call delay for users with packet call arrival 9
process. The averaged packet call delay is defined as the ratio of the accumulated 10
delay for all packet calls for the user and the total number of packet calls for the 11
user. The delay for a packet call is defined as in 4. Note that this definition is 12
applicable only to a user with packet call arrival process. 13
7. The scattering plot of data throughput per user vs. the distance from the 14
user’s location to its serving sector. In case of SHO or sector switching, the 15
distance between the user and the closest serving sector shall be used. The data 16
throughput for a user is defined as in 3. 17
8. The scattering plot of packet call throughputs for users with packet call arrival 18
processes vs. the distance from the users’ locations to their serving sectors. In 19
case of SHO or sector switching, the distance between the user and the closest 20
serving sector shall be used. The packet call throughput for a user is defined as in 21
4. 22
9. The scattering plot of averaged packet delay per user vs. the distance from the 23
mobile’s location to its serving sector. In case of SHO or sector switching, the 24
distance between the user and its closest serving sector shall be used. The averaged 25
packet delay per user is defined as in 2. 26
10. The scattering plot of averaged packet call delays for users with packet call 27
arrival processes vs. the distance from the mobiles’ locations to their serving 28
sectors. In case of SHO or sector switching, the distance between the user and its 29
closest serving sector shall be used. The averaged packet call delay per user is 30
defined as in 4. 31
11. The scattering plot of data throughput per user vs. its averaged packet delay. 32
The data throughput and averaged packet delay per user are defined as in 3 and 2, 33
respectively. 34
12. The scattering plot of packet call throughputs for users with packet call arrival 35
processes vs. their averaged packet call delays. The packet call throughput and 36
averaged packet call delay per user are defined as in 4. 37
13. Applicable to 1xEV-DO BCMCS only: The user coverage probability at 1% PER for 38
each channel model, and for the channel mix specified in Table 2.2.2-1. 39
14. Applicable to 1xEV-DO BCMCS only: For proposals that use a concatenated 40
coding scheme (such as a turbo inner code and a Reed Solomon outer code), the 41
3GPP2 C.R1002-A v1.0
37
packet erasure rate at the input and output of the outer code, for each mobile in the 1
simulation. 2
Appendix D provides formulas of the above definitions. 3
The channel model and speed of a data user are randomly chosen according to the pre-4
determined distributions specified in 2.2.2. 5
2.3.5.3 1xEV-DV Systems Only 6
2.3.5.3.1 Voice Services and Related Output Matrices 7
The following statistics related to voice traffic shall be generated and included in the 8
evaluation report. 9
1. Voice capacity, where the voice capacity is defined as the maximum number of voice 10
users that the system can support within a sector with certain maximum system 11
outage probability. The details on how to determine the voice capacity of a sector 12
are described in Appendix C. 13
2. The histogram of voice data rates (for a frame) per user and for all users. 14
3. The scattering plot of the outage probability vs. the distance from the mobile to the 15
serving cell. In case of soft hand-off (SHO), the distance from the mobile to the 16
closest serving cell shall be used. 17
4. The curve of outage indicator vs. time for each voice user. The outage indicator 18
equals to one when the voice user is in outage, and zero otherwise. The speed, 19
channel model and the distance of the voice user to the serving cell shall also be 20
included in the curve. In case of SHO, the distance from the mobile to the closest 21
serving cell shall be used. 22
5. The outage probability for each user. Note that this value can be calculated from the 23
curve described in the previous item. 24
The channel model and speed of a voice user are randomly chosen according to the pre-25
determined distributions specified in Table 2.2.2-1. 26
2.3.5.3.2 Mixed Voice and Data Services 27
In order to fully evaluate the performance of a proposal with mixed data and voice services, 28
simulations shall be repeated with different loads of voice users. The following outputs shall 29
be generated and included in the evaluation report. 30
1. The following cases shall be simulated: no voice users (i.e., data only), voice users 31
only (i.e., the number of voice users equals to voice capacity), and average 0.5Nmax 32
and 0.8Nmax voice users per sector. 33
2. For each of the above case, all corresponding output matrices defined for voice and 34
data services shall be generated, whenever they are applicable. 35
In addition to the output matrices described in the previous two sections, the following 36
output matrix shall also be generated and included in the evaluation report. 37
3GPP2 C.R1002-A v1.0
38
1. A curve of cell/sector data throughput vs. the number of voice users, where the 1
cell/sector data throughput is defined as above. 2
2.3.5.4 1xEV-DO Systems Only (Mixed Rev. 0 and Rev. A Mobiles) 3
For proposals that support backward compatibility with 1xEV-DO (Rev. 0) ATs, the 4
following mixed Rev. 0 and Rev. A user tests shall be simulated using the standard system 5
layout and channel mix specified in section 2.2.2: 6
a. 8 full-buffer Rev. 0 ATs and 8 full-buffer Rev. A ATs per sector. 7
b. 16 Rev. 0 full-buffer ATs per sector. 8
c. 16 Rev. A full-buffer ATs per sector. 9
The statistics required by the output matrices specified in appendix M shall be generated 10
and presented for the purpose of evaluation, as well as the associated statistics required for 11
the evaluation report specified in 2.3.5.2. In addition, the following statistic shall be 12
reported: 13
a. Avg_TP_Rev0_Mix: Average throughput of Rev. 0 ATs in scenario a. 14
b. Avg_TP_RevA_Mix: Average throughput of Rev. A ATs in scenario a. 15
c. )_Re__
_Re__,
_0Re__
_0Re__(
OnlyvATPAvg
MixvATPAvg
OnlyvTPAvg
MixvTPAvg: 16
where Avg_TP_Rev0_Only denotes the average throughput per AT from scenario b, 17
Avg_TP_Rev_A_Only denotes the average throughput per AT from scenario c. 18
2.3.5.5 UMB Systems Only 19
The default bandwidth for the evaluation with full buffer traffic model is 5 MHz. The 20
simulation shall be performed with the following number of users: 21
a. Number of full buffer traffic users for 5 MHz – 1, 4, 8, 16, 32 user(s)/sector 22
b. Number of full buffer traffic users for 10 MHz – 1, 4, 8, 32 user(s)/sector 23
For the evaluation with mixed traffic model, the following user configuration shall be used: 24
a. 100 VoIP users/sector + 32 full buffer users/sector for 5MHz 25
b. 16 VT users/sector + 16 full buffer users/sector for 5MHz 26
27
2.4 Calibration Requirements 28
Proponents and evaluators of any proposed design should follow the following calibration 29
steps to ensure a consistent evaluation process. 30
2.4.1 Link Level Calibration 31
All link-level data (in the form of Excel spreadsheet) should be submitted. One set of data 32
for use by all evaluators. 33
3GPP2 C.R1002-A v1.0
39
2.4.2 System Level Calibration 1
Each evaluator of the proposal(s) should submit the following calibration data for the 2
simulator. 3
2.4.2.1 UMB System Calibration 4
The following cases shall be simulated in the forward link of UMB system. 5
a. Full buffer traffic only – 10AT/sector, 10MHz 6
AWGN 7
Ped-B 3km/h and Veh-A 120km/h 8
SCM (2 X 2 SCW) 9
b. VoIP traffic only – 100AT/sector, 5MHz 10
Ped-B 3km/h and Veh-A 120km/h 11
Layout files for AN and AT are used[43]. The pathloss in the layout files contains 12
propagation loss, shadowing, and antenna pattern, but does not contain AN and AT 13
antenna gains as well as "other losses" (in total 4dB). Hence, 4 dB should be subtracted 14
from every path-loss value in the layout file. 15
A source file is used for VoIP traffic generation[42]. Maximum C/I of 28dB is used. 16
Reference link curve for FER prediction is used[44]. 8-tile curves are used for full buffer 17
simulation and 1-tile curves are used for VoIP simulation. 18
Details of parameters used in the calibration are summarized in [45]. 19
Output metrics used in the calibration follows the format of [46]. 20
2.4.2.1.1 Scheduler 21
A simplified scheduler is used in the calibration of UMB system simulators. The scheduler 22
is described for the following cases. 23
a. Full buffer users only 24
b. VoIP users only 25
For the calibration purpose, closed loop precoding and/or subband scheduling is not 26
considered. LAB assignment errors are also ignored. Extended transmissions are not 27
supported, even for low geometry users. In the case of VoIP traffic, multiple tile 28
assignments are not supported. 29
The names of functions and variables are usually detailed enough to represent their 30
functionality. 31
Full buffer users only scheduler 32
In forward link full buffer users only case, since subband scheduling is not used, in each 33
frame, the scheduler will assign the entire frame to a single user. This fact can be used to 34
simplify the scheduler. Following provides the pseudo-code for the scheduler used for FL 35
full buffer users only case. Note that all assignments are persistent, i.e., the assignment 36
3GPP2 C.R1002-A v1.0
40
will remain in effect unless it is de-assigned, or there is a decoding failure. No additional 1
LAB is needed to keep the persistent assignment. 2
//Constants:
MaxNumLABs; //Maximum number of LABs allowed in a frame
N; //Number of ATs in the sector
numInterlace; //Number of interlaces (8)
//Scheduler function for full buffer users only
SchedulerFullBufferFL(){
//Check existing persistent assignment. If there is a decoding error, the assignment is assumed
//lost.
UpdateExistingPersistentAssignments();
//Update channel desirability
for (i=1,…,N){
Priority[i]=CQI[i]/(AverageCQI[i] * ResourceAssignedToAT[i]);
//Note that CQI here shall be interpreted as the spectral efficiency (b/s/Hz) translated from
//CQI reports. AverageCQI is the average
//of the previous spectral efficiency using an one-tap IIR filter with time constant 0.6 second.
//The AverageCQI is updated whenever there is a CQI report
//ResourceAssignedToAT is the actual time-frequency resource assigned to the AT
//counted from the beginning of the simulation.
}
//Count the number of LABs used, including RLABs (and possibly those used by VoIP users)
numLABs=NumUsedLABs();
//Find usable channel nodes (not used in active HARQ transmission)
usableChannelNodeList=FindUsableNodes(); //If not empty, the list shall contain a single
node.
if ( (usableChannelNodeList not empty) &&
(numLABs < MaxNumLABs) ){
numLABs++;
//Select user to be assigned
idx=argmax(Priority[i]);
//Assign the node to the AT
AssignNodeToAT(idx, usableChannelNodeList[1]);
//Compute the power density
powerDensity=ComputePowerDensity();
assignedPacketFormat = ComputePacketFormat( powerDensity,
CQI[idx], targetTermination);
//Send LAB
SendLAB(idx, usableChannelNodeList[1], assignedPacketFormat);
}
//All persistent assignment not affected by this function are still effective.}
3
VoIP users only scheduler 4
3GPP2 C.R1002-A v1.0
41
For VoIP user, each AT is assigned one tile (base node) in one of the eight interlaces at a 1
time. 2
All VoIP assignments are persistent. When assigned, the tile-interlace pair is reserved for 3
the AT till it is explicitly or in-explicitly de-assigned. A VoIP assignment will not be explicitly 4
de-assigned in the system simulation. The tile-interlace pair assigned to an AT can be re-5
assigned to another AT without informing the first AT. We assume the first AT will keep 6
decoding and detect a decoding error after 6 HARQ transmissions and release the 7
persistent assignment, but this will not be modeled in the simulator. 8
The priority of VoIP user is decided by the time stamp of the packets in the buffer. The VoIP 9
user with the earlier time stamp has higher priority to be scheduled. 10
When there is no more empty tile-interlace pair available, an assigned AT can be switched 11
out if there is no new packet arrives for the AT for the recent 25ms. 12
Power boost is not applied for HARQ re-transmissions for calibration purpose. 13
For VoIP traffic transmission, two packet formats are used and are optimized for EVRC 14
vocoder. Packet format 4 is used for full-rate and half-rate. Packet format 2 is used for 15
quarter-rate and 1/8-rate. 16
The scheduling algorithm is summarized as follows: 17
18
//Constants:
MaxNumLABs; //Maximum number of LABs allowed in a frame
frameInterval; //Time duration of a frame
N; //Number of ATs in the sector
numInterlace; //Number of interlaces (8)
//Scheduler function for VoIP users only
SchedulerVoIPFL(){
//Check existing persistent assignment. If there is a decoding error, the assignment is assumed
//lost.
UpdateExistingPersistentAssignments();
//Check packet arrival for each AT and update timer
for (i=1,…,N){
if (PacketArrivesForAT(i)){
timer[i].Reset();
AddPacketToBufferWithTimeStamp(i);
}else{
timer[i]+=frameInterval;
}
//Expires old packets
For (i=1,...,N){
ExpirePacketsForAT(i);
}
//Find available channel nodes
3GPP2 C.R1002-A v1.0
42
usableBaseNodeList = FindUsableNodes(timer);
//Find the usable channel nodes. Return a list of base nodes. In the list, the
//empty base nodes will be in the beginning. The releasable base nodes
//(persistently assigned to VoIP AT but not used in the frame and no new
//packet arrives for the AT in the recent 25ms). The releasable base nodes
//are listed in the order of last packet arrive time (earliest first).
//Find ATs to be scheduled in the frame
ToBeScheduled=FindATList(buffer[1,…,N], assigned[1,…,N]);
//Find the list of AT that needs to be scheduled. Generate a list of ATs
//with no assigned resources. The output is ordered in decreasing order
//of the time stamp of the first packet in the buffer.
//Main loop to schedule VoIP ATs
numLABs = LABinRL();
while ( (usableBaseNodeList not empty) &&
(numLABs < MaxNumLABs) &&
(ToBeScheduled not empty)){
//Pick the first AT in the ToBeScheduled list and remove it from the list
idx=ToBeScheduled.PopFirst();
//Pick the first base node in the usableBaseNodeList
node=usableBaseNodeList.PopFirst();
numLABs++;
AssignNodeToAT(node, currentInterlace);
SendLAB(idx, node);
}
//Assign packet formats to all scheduled ATs, including the ones scheduled before.
for (i=1,…,N){
UpdatePacketFormatForAssignment(i);
//Only update the packet format if the AT is not in an active HARQ
//transmission. The packet format is decided by the buffer size. The
//transmissions are power controlled to achieve a target termination point.
}
//All persistent assignment not affected by this function are still effective.
}
1
. 2
3GPP2 C.R1002-A v1.0
43
3 EVALUATION METHODOLOGY FOR THE REVERSE LINK 1
3.1 System Level Setup 2
3.1.1 Antenna Pattern 3
Antenna pattern shall be as specified in 2.1.1. 4
3.1.2 System Level Assumptions 5
The parameters used in the simulation are listed in Table 3.1.2-1. Where values are not 6
shown, the values and assumptions used in the simulation shall be specified in the 7
simulation description. 8
3GPP2 C.R1002-A v1.0
44
Table 3.1.2-1 Reverse Link System Level Simulation Parameters 1
Parameter Value Comments
Number of 3-sector Cells 19 2 ring, 3-sector system, 57
sectors total. These cells are
on a ―wrap-around‖ model
where the signal or
interference from any MS to
a given cell is treated as if
that MS is in the first 2 rings
of neighboring cells. MSs
are uniformly dropped over
the 19 cells. Simulation is
done with the desired
number of MSs for each
sector of each cell.
Throughput and capacity are
collected from all cells
Antenna Horizontal Pattern 70 degree (-3 dB)
with 20 dB front-to-back
ratio
see Section 2.1.1
Antenna Orientation 0 degree horizontal azimuth
is East (main lobe)
No loss is assumed on the
vertical azimuth. (See
Appendix B)
Propagation Model
(BTS Ant Ht=32m, MS=1.5m)
28.6+ 35log10(d) dB,
d in meters
Modified Hata Urban Prop.
Model @1.9GHz (COST 231).
Minimum of 35 meters
separation between MS and
BS. 7
7 In this document the word ―modified‖ represents a difference from the COST231-Hata model
wherein the path loss has been reduced by 3 dB [33]. If a mobile is dropped within 35 meters of a
base station, it shall be redropped until it is outside the 35-meter circle.
3GPP2 C.R1002-A v1.0
45
Parameter Value Comments
Log-Normal Shadowing Standard Deviation = 8.9 dB
for both FL and RL
Independently generate
lognormal per mobile-sector
pair and use the method
described in Appendix A.
This shadowing is constant
in each simulation run. The
same shadowing amount
shall be used for all Rx
antennas of a BS (up to six)
to a given MS. The
correlation coefficient
between the BS‘s Tx
antennas and a given MS
and the BS‘s RX antennas
and a given MS is 1.
Maximum RL Total Path
Loss
146 dB This term includes the MS
and BS antenna gains, cable
and connector losses, other
losses, and shadowing, but
not fading.
Base Station Shadowing
Correlation
0.5 See Appendix A
Overhead Channel Reverse
Link Power Usage
Any additional overhead
needed to support other
control channels (dedicated
or common) for the forward
link or the reverse link must
be specified and accounted
for in the simulation
Base Noise Figure 5.0 dB
Thermal Noise Density -174 dBm/Hz
Carrier Frequency 2 GHz
BS Antenna Gain w Cable
Loss
15 dB 17 dB BS antenna gain; 2
dB cable loss
MS Antenna Gain -1 dBi
Other Losses 10 dB Applicable to all fading
models
Maximum MS EIRP 23 dBm
3GPP2 C.R1002-A v1.0
46
Parameter Value Comments
Fast Fading Model Based on Speed The fading processes on the
paths from a given MS to the
two BS antennas are
mutually independent. The
fading model is specified in
Appendix K.
Active Set
Membership
Circuit
switched and
packet
switched
data systems
(e.g., 1xEV-
DV)
Up to 3 members are in the
Active Set if the pilot Ec/Io
is larger than T_ADD = -18
dB (=9 dB below the FL pilot
Ec/Ior) based on the FL
evaluation methodology
Packet
switched
data systems
(e.g., 1xEV-
DO)
Up to 3 members are in the
Active Set if the pilot Ec/Io
is larger than T_ADD = -9 dB
based on the FL evaluation
methodology
Delay Spread Model See Table 2.2.2-1 and Table
2.2.2-2
Reverse Link Scheduling System specific. Proponents
need to declare the scheme
and the associated MAC
delay and reliability.
Active Set Change System specific. Proponents
need to declare the scheme
and the associated signaling
delay and reliability.
3GPP2 C.R1002-A v1.0
47
Parameter Value Comments
Reverse Link
Power
Control
Circuit
switched and
packet
switched
data systems
(e.g., 1xEV-
DV)
Closed-loop power control
delay: two PCGs8
Update Rate: Dependent on
proposal.
Power control feedback: BER
= 4% for a BS-MS pair.
Different values shall be
specified and accounted for
in the simulation
Ec/Nt measurement error at
the BS: additive in dB, log
normal, zero-mean random
variable with a 2 dB
standard deviation.
Packet
switched
data systems
(e.g., 1xEV-
DO)
Closed-loop power control
delay: two slots8
Update Rate: Up to 600 Hz,
dependent on proposal.
Power control feedback: BER
= 4% for a BS-MS pair.
Different values shall be
specified and accounted for
in the simulation
Ec/Nt measurement error at
the BS: additive in dB, log
normal, zero-mean random
variable with a 2 dB
standard deviation.
MS PA Size 200 mW
Site to Site distance 2.5 km
2.0 km Default site to site distance
for UMB
8 The MS transmit power changes in PCG/slot i+2 in response to measurement made in PCG/slot i.
One PCG/slot delay for link level modeling (measured from the last chip that the reverse pilot is
measured to the time that the mobile changes TX power level).
3GPP2 C.R1002-A v1.0
48
Parameter Value Comments
Rise over Thermal (Reverse
Received Power Normalized
by Thermal Noise Level)
7 dB Histogram of this parameter
with a 1.25 (1xEV-DV) or
1.67 (1xEV-DO) ms time
resolution shall be provided
with the mean rise-over-
thermal. The percentage of
time the rise over thermal
above the 7 dB target shall
not exceed 1%. Rise over
thermal for the default two
receiving antenna mode is
½ [(Io1+No)/No + (Io2 +
No)/No], where the total
received signal power at
antenna i is defined as Ioi,
i=1,2.
1
3.1.3 Call Setup Model 2
The following is the method to simulate a call setup on the RL, regardless of the traffic time: 3
1. MS gets in the system at t = t09 4
2. At t = t0, MS starts transmitting pilot only for 320 ms10 or 427 ms10 for 1xEV-DV or 5
1xEV-DO systems respectively 6
a. Closed loop power control is active but outer loop has a fixed target pilot 7
Ec/Io 8
b. Starting pilot transmission power is –50 dBm 9
c. BS Set Point is fixed at – 22.5 dB or –20.5 dB for 1xEV-DV or 1xEV-DO 10
systems respectively 11
3. At t = t0 + 320 ms, together with the pilot, MS begins transmission of the null traffic 12
at 1.5 kbps using the traffic to pilot ratio of -5.875 (= -47/8) dB for 1xEV-DV 13
systems. At t = t0 + 427 ms, together with the pilot, the MS starts the DRC 14
transmission using a specified DRC-to-pilot ratio for 1xEV-DO systems. (For 1xEV-15
9 t0 is a relative time, i.e. it is not the absolute system time but the moment when a MS gets in the
system. For HTTP and WAP users, t0 is at the very beginning of the simulation, while for FTP upload
MS, t0 occurs during the simulation, as these MSs arrive in the system according to the Poisson
arrival process.
10 This step is meant to replace the probing process. Although it does not represent the real system
events, it approximately models the adjustment of MS transmit power, delay and loading on RL
3GPP2 C.R1002-A v1.0
49
DO Rev. 0, the DRC-to-pilot ratio equals to -1.5 dB for non-handoff ATs (DRC 1
Length = 2 slots) and –3 dB for ATs in soft-handoff (DRC length = 4 slots).) 2
a. Outer loop power control is active 3
4. At t = t0 + 580 ms, call setup procedure is finished 4
a. Statistic collection for the MS starts at the end of the call setup 5
b. MS can now start using R-FCH and request R-SCH 6
The timeline diagram is shown in Figure 3.1.3-1 below. 7
8
t = t0
t = t0 + 320ms t = t
0+ 580ms
Pilot
Null Traffic
Closed-loop power control,
Fixed outer loop setpointClosed-loop power control
+ Outer loop power control
Regular Traffic
9
Figure 3.1.3-1: Simplified Call Setup Timeline for 1xEV-DV. (Timeline for 1xEV-DO is 10
the same by modifying 320 ms to 427 ms) 11
3.1.4 Packet Scheduler 12
The voice users‘ (when simulated together with the data users) transmissions are not 13
scheduled in 1xEV-DV systems. The data users can be scheduled or allowed to transmit in 14
a random fashion. The exact procedure, and its delay and reliability, with which a mobile 15
station gains the right to transmit, shall be specified in detail. Proponents shall state 16
whether the reverse scheduling signaling is sent from the entire Active Set of the MS or its 17
subset. As a baseline, the scheduler shall be unaware of the application(s) running on the 18
MS‘s and shall not use the information provided by the QoS BLOB. Baseline simulation 19
results shall be generated accordingly and be presented. The proponent may, however, 20
present additional results with a more sophisticated scheduler in the system. 21
3.1.5 Backhaul Overhead Modeling in Reverse Link System Simulations 22
The backhaul bandwidth used by signaling and measurement messages is provided in the 23
format shown in Table 3.1.5-1. It is assumed that the messages sent over the backhaul 24
links use the TCP/IP protocol. Therefore, a TCP/IP packet overhead of 320 bits (40 bytes) is 25
added to each signaling or measurement message sent over the backhaul. The backhaul 26
overhead on the FL is given by: 27
b/s S1
overhead Backhaul FL 1
i
FLN
irunT 3.1-1 28
Where Si is the signaling message size in bits (including 320 bits TCP/IP header) and NFL is 29
the total number of messages (signaling, measurements etc.) sent on the FL during a 30
simulation run. Trun is the simulation time in seconds for a given run. The FL backhaul 31
overhead is averaged over all the simulation runs. 32
3GPP2 C.R1002-A v1.0
50
The backhaul overhead on the RL is given by: 1
b/s S1
overhead Backhaul RL 1
j
RLN
jrunT 3.1-2 2
Where Sj is the signaling message size in bits (including 320 bits TCP/IP header) and NRL is 3
the total number of messages sent on the RL during a simulation run. The RL backhaul 4
overhead is averaged over all the simulation runs. 5
Table 3.1.5-1 Backhaul bandwidth used by signaling and measurement messages 6
FL backhaul overhead [b/s] RL backhaul overhead [b/s]
Under study Under study
3.1.6 Simulation of Forward Link Overheads for Reverse Link System Simulation 7
For reverse-link enhancement proposal envisioning communicating schedule grants (rate 8
assignments), rate adjustments, acknowledgements (to support H-ARQ operation) from one 9
or more base station(s) to the data mobile, on new or existing Forward Link channels, the 10
following shall apply when modeling the impact of these forward-link (FL) channels on RL 11
performance: 12
1. The impact of imperfect RPC on reverse-link power-control shall be modeled by 13
assuming a 4% error rate on the RPC channel for each BS-MS pair. The closed-14
loop power-control delay shall be 1 slot/PCG. 15
2. For proposals with hybrid-ARQ scheme, the impact of imperfect FL ARQ signal 16
shall be modeled in the network simulation with dynamic FL modeling in the 17
simulations. The dynamic FL modeling shall assume the same multipath and 18
Doppler channel model on FL and RL (i.e. channel A, B, C, D, and E). The 19
fading process on FL and RL shall be assumed as independent. 20
3. Any additional FL channel whose error performance may impact the 21
performance on RL shall be modeled either via dynamic FL modeling (assuming 22
the same multipath and Doppler channel model but independent fading on FL 23
and RL) or via static modeling, both of which are described below. 24
Both static and dynamic simulation methods are allowed for modeling these Forward Link 25
Overhead channels that support Reverse Link operation.11 Static simulation results are 26
required to be provided; dynamic simulation results may be provided. These two methods 27
are described below. 28
3.1.6.1 Static Modeling Method 29
1. Long-term (FER versus average Eb/No) error curves are to be generated, over each 30
channel model, for each of the proposed overhead channels12 on the Forward Link. 31
11 If channel sensitive scheduling is used on the forward link, the static method may be pessimistic.
12 These curves are generated for each frame format allowed for the overhead channels.
3GPP2 C.R1002-A v1.0
51
2. For users receiving the overhead channels in handoff, long term error curves are to 1
be generated for each channel model, way of soft hand-off (SHO), and a range of 2
SHO imbalance(s), and Geometry. 3
3. A desired target FER shall be specified for each of these channels for proper system 4
operation. 5
4. For each user dropped into a system simulation run, the overhead channel frame 6
format is determined, the Geometry is determined, and the Eb/No required to meet 7
the target FER is translated into an average fractional power allocation at each 8
serving base station. Each mobile is assigned the same channel model on the 9
Forward Link as on the Reverse Link. 10
The following methods apply in addition to the above to 1xEV-DV systems only: 11
1. The fractional power allocations for all the users in each drop are summed at each 12
base station. This is the average fractional power cost on the forward link, for that 13
drop at each base station. 14
2. Average these fractional power cost values across the required number of RL system 15
simulation drops, at each base station. This is the average power cost on the 16
forward link at each base station. 17
3. Similarly for each drop, determine the number of Walsh channels used to support 18
the reverse link at each base station. Average the number of Walsh channels used 19
across the required number of RL system simulation drops, at each base station. 20
This is the average number of Walsh channels used on the forward link at each base 21
station. 22
Reverse Link System Simulation Assumptions and Modeling 23
Independent error events should be generated in the Reverse Link system simulation at 24
time instants corresponding to the reception of any Forward Link overhead channel. An 25
error shall occur on each reception with a probability equal to the specified target FER for 26
the corresponding channel. The resulting impact (schedule grant miss, error in rate 27
assignment, ACK/NACK feedback error, etc.) shall be modeled and modeling mechanism 28
specified. 29
3.1.6.2 Dynamic Modeling Method 30
The dynamic modeling of Forward Link overheads is done by running fading on the 31
Forward Link (operating only the Forward Link overhead channels) in conjunction with the 32
Reverse Link in the Reverse Link system simulation. 33
1. Short-term (FER Vs average Eb/No) error curves are to be generated, over each 34
channel model, for each of the proposed overhead channels on the Forward Link. 35
2. For users receiving the overhead channels in handoff, short term error curves are to 36
be generated for each channel model, way of soft hand-off (SHO), and a range of 37
SHO imbalance(s), and instantaneous Geometry. 38
3. A desired target FER shall be specified for each of these channels for proper system 39
operation. 40
3GPP2 C.R1002-A v1.0
52
4. The Forward Link fades independently of the Reverse Link at each connected base 1
station. A connected base station is one that sends schedule grants, rate 2
adjustments, or acknowledgements to a mobile. Each mobile is assigned the same 3
channel model on the Forward Link as on the Reverse Link. 4
5. Power allocation for each of the overhead channels, in the system simulation, is 5
done dynamically and logged. The power allocations should be done for all 6
members of the active set. The proponent of the proposal should explain how the 7
power allocations are done. [For example, for the serving base station, the CQI 8
value is fed back to the base station by the mobile station and the forward link 9
power is set based upon the returned CQI value.] 10
The following apply in addition to the above for 1xEV-DV systems only: 11
1. The instantaneous power allocations for each user are averaged over the drop for 12
every base station. Then these average power allocations are summed over all users 13
at each base station. This is the average fractional power cost on the forward link, 14
for that drop at each base station. 15
2. Average these fractional power cost values across the required number of RL system 16
simulation drops, at each base station. This is the average power cost on the 17
forward link at each base station. 18
3. Similarly for each drop, determine the number of Walsh channels used to support 19
the reverse link at each base station. Average the number of Walsh channels used 20
across the required number of RL system simulation drops, at each base station. 21
This is the average number of Walsh channels used on the forward link at each base 22
station. 23
3.1.6.3 Quantification of Forward Link Overhead as a Data Rate Cost (1xEV-DV Systems 24
Only) 25
An appropriate metric is needed for specifying the cost (in power and bandwidth) on the 26
Forward Link associated with the Forward Link overhead channels supporting Reverse Link 27
operation. For both the static and dynamic methods, the average amount of forward link 28
power and the number of required Walsh channels are determined. 29
The Forward Link (Revision C) system simulations are run with a new amount of overhead 30
power and a new number of available Walsh channels. Specifically, the amount of 31
overhead power is increased by the average amount of overhead power determined from 32
either the static or dynamic system level simulation. Similarly the number of available 33
Walsh channels is reduced by the number of Walsh channels used to support the reverse 34
link as determined by either the static or dynamic method. In determining the resulting 35
capacity reduction, 40 forward link users should be used in a data-only simulation. 36
3.1.7 Signaling Errors 37
Signaling errors shall be as specified in section 2.1.5. 38
For modelling the signalling error in TCP three-way handshake protocol as shown in Figure 39
4.2.2-1 in FTP upload traffic model, failed RL handshake packets will be re-transmitted in 40
3GPP2 C.R1002-A v1.0
53
the physical layer up to a number of times per proposal, if physical layer ARQ is used. If it 1
still fails after physical layer retransmission, it will be assumed error free from TCP protocol 2
viewpoint, but the error is modeled from throughput counting viewpoint. The handshake 3
and ACK packets on the FL are assumed to be error free. However, the delay element of the 4
FL handshake and ACK packets is modeled. 5
3.1.8 Fairness Criteria 6
The CDF of the normalized throughputs with respect to the average user throughput for all 7
FTP upload MS in the same sector is determined. This CDF shall lie to the right of the curve 8
given by the three points in Table 3.1.8-1. 9
Table 3.1.8-1 CDF Criterion for FTP Upload MS 10
Normalized Throughput w.r.t
average user throughput CDF
0.1 0.1
0.2 0.2
0.5 0.5
The fairness criterion with geometric mean and harmonic mean that is described in 2.1.6.2 11
will be used for the evaluation of full buffer MS in UMBsystem. 12
3.1.9 FER Criterion 13
The following outage criterion is defined for the final FER after physical layer 14
retransmissions: 15
For each traffic channel type (e.g., R-FCH, R-DCCH, R-SCH, RTC or other new RL 16
traffic channels proposed), the mean FER across all users in the system shall be 17
sufficient to sustain services. 18
For each traffic channel type (e.g., R-FCH, R-DCCH, R-SCH, RTC or other new RL 19
traffic channels proposed), the percentage of users with FER greater than 5% shall 20
be low enough to insure adequate service delivery and coverage. 21
3.1.10 IoT Criterion 22
The Interference over Thermal (IoT) criterion will be used for the simulation of UMB system. 23
The IoT is computed per sector ‗s‘ per slot and is defined as 24
0
)(:
0,,
N
NE
IoTsiBSi
sic
s
3.1-3 25
where BS(i) represents the best serving sector of access terminal i, sicE ,, is the total received 26
power from access terminal i, at sector s and 0N is the thermal noise variance. 27
3GPP2 C.R1002-A v1.0
54
The IoT statistics are collected from all sectors of the system. The reverse link operation 1
shall use an IoT criterion of a mean value of 6 dB and any value larger than 6dB such that 2
the other performance criteria are not violated. The IoT for the default two receiving 3
antenna mode is ½ [(IoT1+No)/No + (IoT2 + No)/No], where the total received signal power at 4
antenna i is defined as IoTi, i=1,2. 5
3.2 Link Level Modeling 6
3.2.1 Link Level Parameters and Assumptions 7
The performance characteristics of individual links used in the system simulation are 8
generated a priori from link level simulations. Link level simulation parameters are 9
specified in Appendix J. 10
Turbo Decoder Metric and Soft Value Generation into Turbo Decoder shall be as specified 11
in Appendix H. 12
3.2.1.1 Frame Erasures 13
Data for the link-level performance shall be presented and agreed upon so that one set of 14
data is used in all simulations. 15
The following method describes the Quasi-Static approach for modeling link performance. 16
Additional link level modeling methods can be found in Appendix N and Appendix O: 17
Equivalent SNR Method based on Convex Metric (ECM) described in Appendix O shall be 18
used to generate frame erasures for UMB system. Justification on the parameter values 19
used shall be provided by each proponent. The link level statistics that are used for 20
generating the short-term FER curves for link-to-system mapping is shown in Appendix W. 21
Description of Quasi-Static Approach with Short Term FER: 22
The quasi-static approach with short term FER shall be used to generate the frame 23
erasures for reverse link channels with fixed traffic to pilot power (T/P) ratio in a frame 24
such as R-FCH, R-DCCH, R-SCH, RTC, other new proposed RL data channels, and RL 25
overhead channels. 26
A full set of short term FER vs. average traffic Eb/Nt per frame curves is generated as a 27
function of radio configurations, data rates, T/P ratios, and channel models. The number of 28
curves should be reduced if possible, provided that this won‘t unduly affect the validity of 29
this quasi-static approach. 30
In the system-level simulation, the average traffic Eb/Nt per frame is computed as follows. 31
First, the traffic Eb/Nt is calculated in a PCG/slot as specified in section 3.2.1.3. The short-32
term average traffic Eb/Nt per frame is defined as the average of the traffic Eb/Nt for all 33
PCGN PCGs/slots in a frame, i.e., 34
PCGN
n nt
b
PCG
tbN
E
NNEAvg
1
1)/( 3.2-1 35
where ntb NE )/( is the traffic Eb/Nt in the n-th PCG in a frame. Once the )/( tb NEAvg is 36
calculated as in the above equation, it is used to look up the corresponding link level short 37
3GPP2 C.R1002-A v1.0
55
term FER vs. average Eb/Nt per frame curve for the specific condition. A frame erasure 1
event is then generated based on the FER value. 2
In HARQ scheme, a full set of short term FER vs. total traffic Eb/Nt curves is generated as a 3
function of the number of transmission frames for the specific condition. The total traffic 4
Eb/Nt is defined as the summation of )/( tb NEAvg of transmitted frames. It is used to look 5
up the corresponding link level short term FER vs. total traffic Eb/Nt curve for the number 6
of transmitted frames. 7
The short-term FER curves shall be generated as follows: 8
1. The link-level simulation is conducted for the same condition as in system level 9
simulation as much as possible. The average traffic Eb/Nt in a frame and the frame 10
erasure indicator for the frame are recorded. The average traffic Eb/Nt per frame is 11
computed as follows in the link-level simulation 12
PCG S ym b o lN
n
N
k
a j
jaknt
jaknp
a j
jaknpc
SymbolPCGtb
NE
PGEE
NNNEAvg
1 1 ),,,(2
),,,(
22
),,,(
1)/(
3.2-2 13
where n is the index of PCG/slot in a frame, k is the index of symbols within a 14
PCG/slot, NSymbol is the number of symbols in a PCG/slot, a is the index of antennas 15
in a receiver, and j is the index of paths in an antenna. cE is the transmitted chip 16
energy of traffic channel, pE is the transmitted chip energy of pilot channel, and 17
PG is the processing gain defined by the ratio of traffic channel data rate to chip 18
rate. ),,,( jakn is the attenuation factor in the j-th path of the a-th antenna for the k-19
th received coded symbol in the n-th PCG/slot. ),,,( jakntN is the total variance of noise 20
and non-orthogonal interference component in the j-th path of the a-th antenna for 21
the k-th received symbol in the n-th PCG/slot. 22
2. In HARQ scheme, the total traffic Eb/Nt is calculated as the summation of 23
)/( tb NEAvg of transmitted frames. To generate short term FER curve after R-th 24
transmissions of frame, the total traffic Eb/Nt after R-th transmissions and the 25
frame erasure indicator are recorded only when the frame is erased after (R-1)-th 26
transmission. 27
3. Generate the histogram of short term FER, i.e., the range of Eb/Nt is divided into 28
many bins, and the FER in each bin is computed based on the outputs mentioned 29
in step 1. The size of each bin is 0.5 dB. 30
3.2.1.2 Target FER 31
The operating frame erasure rate or FER will be 1% for voice for 1xEV-DV systems. 32
The FER for the data-only mobile stations shall be: 33
1. The final FER after retransmissions should be less than or equal to 1% 34
3GPP2 C.R1002-A v1.0
56
2. For 1xEV-DV systems, if the R-FCH is present, the pilot level shall be at least as high as 1
what is required to obtain 1% FER on a 9.6 kbps channel, Otherwise the pilot level should 2
not be below the minimum searcher requirements and must be explicitly stated. 3
3. The pilot level should not be below the minimum searcher requirements - the pilot level 4
used must be explicitly stated in such cases. 5
4. Error rates on the F-ACK channel should be specified by the proponent of each RL 6
design. The necessary FL cost is to be shown as in section 3.2.2. 7
3.2.1.3 Channel Models 8
Channel models shall be as Table 2.2.2-1 and Table 2.2.2-2 specified in section 2.2.2. 9
When modeling the fading on a path between a MS and a sector that is not in that MS‘s 10
Active Set, an alternative method may be used as follows where a single path is used to 11
replace the channel models in 2.2.1 to speed up the simulation. In this alternative method, 12
For Channel Models A through D, the replacement path has Rayleigh fading with a 13
Doppler speed of the corresponding channel model 14
For Channel Model E, a non-fading replacement path is used. 15
The received traffic Eb/Nt for the i-th mobile station at the base station is given by 16
SL ANT i
SL ANT i
N
s
N
r
J
j
jrsitjrsisi
N
s
N
r
J
j
jrsisiitrafficc
itraffict
b
Ng
gEPG
N
E
1 1 1
,,,,
2
,,,,
2
1 1 1
2
,,,,,_
,
3.2-3 17
o
N
sNk
rskskkc
sNm
imm
rsm
J
n
nrsmsmmcrsi
J
jff
frsisiicjrsit
NgE
gEgEN
MS
AS
AS mi
1)(
2
,,,,
)(
1
2
,,
1
2
,,,,,
2
,,
1
2
,,,,,,,,,
3.2-4 18
where 19
SLN The number of the softer handover legs 20
ANTN The number of receiving antennas 21
)(sN AS The number of MS‘s that communicate with the s-th sector 22
MSN The number of total MS in the system 23
icE , Transmitted total (traffic + pilot) chip energy of the i-th MS 24
itrafficcE ,_ Transmitted traffic chip energy of the i-th MS 25
3GPP2 C.R1002-A v1.0
57
iJ The number of rays used in a particular channel model of the i-th MS, 1
excluding the ray used to model FURP. 2
sig , The average link gain between the i-th MS and the s-th sector. This term 3
includes the MS and BS antenna gains, cable and connector losses, other 4
losses, and shadowing, but not fading. 5
jrsi ,,, Samples of the fading processes of the j-th recovered rays between the i-th 6
MS and the r-th antenna of the s-th sector. It is used only for the MS‘s 7
that have the s-th sector as an active set member. 8
rsi ,, Samples of the fading process between the i-th MS and the r-th antenna 9
of the s-th sector, for the additional ray used to model interference due to 10
the FURP. It is used only for the MS‘s that have the s-th sector as an 11
active set member. 12
rsk ,, Samples of the fading process between the k-th MS and the r-th antenna 13
of the s-th sector. It is used for the MS‘s that do not have the s-th sector 14
as an active set member. For Channel Model E, rsk ,, = 1. 15
0N The power of AWGN noise 16
The received traffic Eb/Nt shall be used to determine the frame erasures. 17
3.2.2 Forward Link Loading 18
The link level overhead channel performance (FER or BER) curves are to be simulated and 19
submitted in the same set of fading models in 2.2.1. In addition, the loading (percentage of 20
BS transmit power and percentage of Walsh space) of these FL channels shall be evaluated 21
and submitted for 1xEV-DV systems. 22
3.2.3 Reverse Link Power Control 23
The proponents of each reverse link proposal shall completely specify the algorithms for the 24
open and closed loop power control if they are an integral part of the link. This includes the 25
specification of how outer loop power control is accomplished. For a complete analysis of 26
the proposal, it is preferable that the power control algorithms form part of the system 27
simulation. If the proponents feel that it is impractical to simulate the power control 28
algorithm as part of their system simulation and instead, model the effects of power control 29
by changing the statistics of the mobile radio channel, the model shall be completely 30
presented. Moreover, the proponents shall provide analysis to justify that their power 31
control algorithm will modify the mobile radio channel statistics to that used in the system 32
simulation. 33
For calibration purposes, the proponent shall simulate the following inner loop power 34
control algorithm in system level simulation. 35
Calculate the estimated received pilot power jrsipE ,,,,
~ for the j-th path between the i-36
th MS and the r-th antenna of the s-th sector, that is defined by 37
3GPP2 C.R1002-A v1.0
58
2,,,,
2
,,,,,_,,,,,,,,
)/5.0(
)/5.0(~
chipjrsitQ
chipjrsitIipilotcsijrsijrsip
NN
NNEgE
3.2-5 1
where ipilotcE ,_ is the transmitted pilot chip energy of the i-th MS. )(xI and )(xQ 2
are independent Gaussian random numbers with zero mean and standard deviation x . 3
chipN is the measurement duration in chips (i.e., 2048 chips for 1-slot duration in 1xEV-4
DO systems or for 1xEV-DV systems, 1152 in case of PCB puncturing and 1536 in case 5
of no PCB puncturing in the reverse pilot channel). Other notations are specified in 6
section 3.2.1.3. 7
Calculate the estimated total received pilot power, that is defined by 8
SL ANT iN
s
N
r
J
j
jrsitjrsipipilottc NENE1 1 1
,,,,,,,,,)/
~(/
~ 3.2-6 9
Compare the ipilottc NE,
/~
with the target set point in order to make power control 10
command. 11
For calibration purposes, the proponents shall simulate the following outer loop power 12
control set point algorithm: 13
Increase the power control set point by up=0.5 dB if the frame is decoded in error. 14
Decrease the power control set point by downdB =up /(1/FER-1) dB, if the frame is 15
correctly decoded 16
where power control set point is the comparison threshold for the received pilot Ec/Nt at a 17
base station, and Ec is the received pilot energy and Nt is the total interference and thermal 18
noise. 19
The adjustment of the set point in response to the reception of frame i takes place at 20
the beginning of the frame i+2. 21
For 1xEV-DV systems, the outer loop power control set point algorithm shall be constrained 22
to the following range for voice-only operation: 23
Minimum power control set point shall be –26 dB 24
Maximum power control set point shall be –17 dB. 25
3.3 Simulation Requirements 26
3.3.1 Simulation Flow 27
There are 19 3-sectored cells in the simulated system. All cells are fully simulated and 28
monitored for the throughput or voice capacity. See Appendix I for details. 29
3.3.1.1 Soft and Softer Handoff 30
Calls can be in soft or softer handoff, depending on the path losses and their distributions. 31
When in soft handoff, the target Eb/(Io+No) for a given call is assumed to be the same at all 32
3GPP2 C.R1002-A v1.0
59
the sectors in the Active Set while the actual received Eb/(Io+No) values differ by the path 1
loss difference. For 1xEV-DV systems, the same voice link level curves are to be used in soft 2
handoff for the reverse link as in the non-soft handoff case. One power control data stream 3
including power control errors is to be used in the case of softer handoff. 4
3.3.1.2 Simulation Description 5
1. The system consists of 19 cells, each with an imaginary13 hexagonal coverage area. 6
Each cell has three sectors. 7
2. Mobile stations are uniformly dropped into the 19-cell system.14 A MS dropped 8
within 35 meters of a base station is redropped. One fixed path loss (used for both 9
forward link and reverse link) and one randomized shadowing component (for the 10
forward link and for the reverse link) are computed for each BS-MS pair for the 11
duration of the simulation run. Two independent reverse link fading components 12
are computed once every PCG for the two BS receiving antennas in each sector for 13
each MS. 14
3. The sector with the smallest total path loss is the serving sector of the MS. When 15
evaluating per-sector data (number of MSs, sector throughput, average throughput 16
for MSs in a sector, etc.), the MS is counted towards its serving sector. For 1xEV-17
DV systems, each MS shall be assigned to a sector if the sector is in the Active Set 18
(with a maximum size of three) of the voice user. A sector is in the Active Set of a 19
mobile station only if the FL pilot Ec/Io15 of that sector computed without fading is 20
above T_ADD (in Table 3.1.2-1) at the mobile station. A MS dropped into the system 21
shall be redropped if there are no members in its Active Set or if the total path loss 22
to the serving sector is above a threshold defined in Table 3.1.2-1. All sectors of the 23
system shall continue to accept mobile stations until the desired fixed number of 24
MSs per sector is achieved everywhere. 25
4. Fading signal and fading interference are computed from each mobile station into 26
each sector for each PCG or equivalent power control related time interval. See 27
Section 3.2.1.3. 28
5. All reverse links from every MS to all Active Set members (sectors in communication 29
with the MS) are to be modeled for frame erasure, outage, and throughput. 30
6. Applicable to 1xEV-DV Systems Only: For simulation of systems loaded only with 31
voice-only mobiles, the first run is done with five mobile stations per sector and 32
13 The actual coverage areas are determined by propagation, fading, antenna patterns, and other
factors.
14 Any method that is equivalent to the following method is allowed: For each of the19 possible cells,
the MS is uniformly dropped over an imaginary hexagonal coverage area.
15 The forward link pilot Ec/Io is the ratio of the pilot energy per chip and the sum of thermal noise
density and power spectral density of all forward link sectors in the system, assuming all sectors are
transmitting at full power.
3GPP2 C.R1002-A v1.0
60
every run following that is done with five more mobile stations per sector until the 1
outage criteria (defined in Appendix C) or the rise-over-thermal limit is violated, after 2
which every run is done with one less mobile until the outage criteria and the rise-3
over-thermal limit are satisfied. The maximum number of users per sector of Nmax 4
(voice capacity) is achieved if Nmax + 1 users per sector would not satisfy the outage 5
criteria or rise-over-thermal limit and Nmax users per sector would satisfy both. 6
7. For simulation of the system loaded with data-only mobiles, the runs are done with 7
the desired number of HTTP and WAP MSs in each sector as well as a number of 8
FTP upload users that arrive at the system according to a pseudo-random arrival 9
process. To evaluate the system capacity, the inter-arrival parameter controlling the 10
arrival process is decreased until the stopping criteria in step 9 below are met 11
8. Applicable to 1xEV-DV Systems Only: With voice-only and data-only mobile 12
stations in the same simulated system, the number of voice-only mobile stations is 13
fixed at 0.5Nmax or 0.8Nmax per sector for the center cell. The numbers of HTTP 14
and the WAP MSs per sector are also fixed, as specified in Section 4.2.1.2. 15
Throughput from all data users at all Active Set members is to be counted for each 16
category separately. 17
9. Run-stopping criterion: The increase of FTP upload user arrival rate is stopped 18
when one of the following conditions happens: 19
A. For 1xEV-DV systems, voice-only users in the whole system as a group are in 20
outage according to Appendix C. 21
B. Fairness criteria are violated for FTP. 22
C. Packet call throughput criteria are violated for FTP. 23
D. Delay criteria are violated, in one of the traffic categories: HTTP, WAP, TCP ACK 24
and Gaming. 25
E. FER criterion for data users in the whole system is violated. 26
F. The rise-over-thermal limitation is violated. (For this purpose, the rise-over-27
thermal statistics are collected from all sectors of the system. The RoT outage is 28
computed from those system-wide statistics.). System throughput is valid from 29
simulations runs with none of these possible violations. 30
G. The system becomes unstable. In other words, the average rate of FTP users 31
entering the system is larger than the average rate of FTP users exiting the system. 32
Note: To simplify the simulation, only two-way or three-way handoff is used. A mobile 33
station on a two-way or three-way soft handoff counts as one user on the sector with the 34
smallest total path loss in the Active Set. 35
3.3.1.3 Layout Files 36
A pair of layout files is used for all the network simulations in UMB, one for AN and one for 37
AT layout as specified in Section 2.3.3. 38
39
3GPP2 C.R1002-A v1.0
61
3.3.2 Outputs and Performance Metrics 1
3.3.2.1 General Output Matrices 2
The following matrices shall be generated and included in the evaluation report. 3
1. All link-level results used in system-level simulator. 4
2. The histogram of C/I that a base station observed for all users in system-level 5
simulation, where the C/I includes path loss and shadowing. 6
3. The performance of any estimator(s) or predictor(s) that are required by the 7
proposal. For instance, if a channel predictor is used in the proposal, the 8
proponent(s) shall provide the details of the predictor/estimator, its bias, and 9
standard deviation. 10
4. The proponent shall justify the performance of any reverse channels that are not 11
simulated by the link-level or system-level simulator. For example, if the system-12
level simulator does not simulate the reverse signaling channels, the proponent 13
companies shall specify the performance of these channels and justify their claims 14
(see section 2.1.5). 15
5. In order to evaluate the impact of the proposed reverse or forward link modifications 16
to the system performance, the proponent shall specify all additional resources that 17
are consumed in the forward and reverse links. These resources are not limited to 18
the additional overhead channels needed to support the reverse link. It should also 19
encompass parts of existing channels that are consumed (e.g., layer 3 messaging on 20
an existing channel). Furthermore, the proponent shall present an analysis of the 21
system impact due to the consumption of these resources. [Pending decision on 22
dynamic/static simulation] 23
6. The histogram and CDF of rise-over-thermal in 1.25 ms or 61.6 ms (depending on 24
system) time resolution for all sectors. 25
3.3.2.2 Data Services and Related Output Matrices 26
The following statistics related to data traffics shall be generated and included in the 27
evaluation report 28
1. Data throughput per sector. The data throughput of a sector is defined as the 29
number of information bits per second that a sector can receive successfully from all 30
data users it serves, provided that all data users satisfy certain fairness criterion, 31
including fairness in terms of per user throughput as well as delay, and that certain 32
number of voice users (for 1xEV-DV systems) can be maintained with certain GOS. 33
2. Averaged packet delay per sector. The averaged packet delay per sector is defined 34
as the ratio of the accumulated delay for all packets for all users served by the 35
sector and the total number of packets. The delay for an individual packet is defined 36
as the time between when the packet enters the queue at transmitter and the time 37
when the packet is received successively by the base station. If a packet is not 38
successfully delivered by the end of a run, its ending time is the end of the run. 39
3GPP2 C.R1002-A v1.0
62
3. The histogram of data throughput per user. The throughput of a user is defined 1
as the ratio of the number of information bits that the user successfully delivers 2
during a simulation run and the simulation time. Note that this definition can be 3
applied to all data users. 4
4. The histogram of packet call throughput for users with packet call arrival 5
process. The packet call throughput of a user is defined as the ratio of the total 6
number of information bits that an user successfully delivers and the accumulated 7
delay for all packet calls for the user, where the delay for an individual packet call is 8
defined as the time between when the first packet of the packet call enters the 9
queue at transmitter and the time when the last packet of the packet call is 10
successively received by the receiver. 11
5. The histogram of averaged packet delay per user. The averaged packet delay is 12
defined as the ratio of the accumulated delay for all packets for the user and the 13
total number of packets for the user. The delay for a packet is defined as in 2. Note 14
that this definition is applicable to a data user without packet call arrival process. 15
6. The histogram of averaged packet call delay for users with packet call arrival 16
process. The averaged packet call delay is defined as the ratio of the accumulated 17
delay for all packet calls for the user and the total number of packet calls for the 18
user. The delay for a packet call is defined as in 4. Note that this definition is 19
applicable to a user with packet call arrival process. 20
7. The scattering plot of data throughput per user vs. the distance from the 21
user’s location to its serving sector. In case of SHO or sector switching, the 22
distance between the user and the closest serving sector shall be used. The data 23
throughput for a user is defined as in 3. 24
8. The scattering plot of packet call throughputs for users with packet call arrival 25
processes vs. the distance from the users’ locations to their serving sectors. In 26
case of SHO or sector switching, the distance between the user and the closest 27
serving sector shall be used. The packet call throughput for a user is defined as in 28
4. 29
9. The scattering plot of averaged packet delay per user vs. the distance from the 30
mobile’s location to its serving sector. In case of SHO or sector switching, the 31
distance between the user and its closest serving sector shall be used. The averaged 32
packet delay per user is defined as in 2. 33
10. The scattering plot of averaged packet call delays for users with packet call 34
arrival processes vs. the distance from the mobiles’ locations to their serving 35
sectors. In case of SHO or sector switching, the distance between the user and its 36
closest serving sector shall be used. The averaged packet call delay per user is 37
defined as in 4. 38
11. The scattering plot of data throughput per user vs. its averaged packet delay. 39
The data throughput and averaged packet delay per user are defined as in 3 and 2, 40
respectively. 41
3GPP2 C.R1002-A v1.0
63
12. The scattering plot of packet call throughputs for users with packet call arrival 1
processes vs. their averaged packet call delays. The packet call throughput and 2
averaged packet call delay per user are defined as in 4. 3
In order to understand more easily these definitions, some mathematical formula and 4
figures are provided in Appendix D. 5
The channel model and speed of a data user are randomly chosen according to the pre-6
determined distributions. 7
3.3.2.3 1xEV-DV Systems Only 8
The applicable statistics required by the output matrices specified in Appendix M shall be 9
generated and presented for the purpose of evaluation. In addition, the following statistics 10
shall also be reported. 11
3.3.2.3.1 Voice Services and Related Output Matrices 12
The following statistics related to voice traffics shall be generated and included in the 13
evaluation report. 14
1. Voice capacity. Voice capacity is defined as the maximum number of voice 15
users that the system can support within a sector with certain maximum outage 16
probability. The details on how to determine the voice capacity of a sector are 17
described in Appendix C. 18
2. The histogram of voice data rates (for a frame) per user and for all users. 19
3. The scattering plot of the outage probability vs. the distance from the 20
mobile to the serving sector. In case of soft hand-off (SHO), the distance from 21
the mobile to the closest serving sector shall be used. 22
4. The curve of outage indicator vs. time for each user. The outage indicator 23
equals to one when the voice user is in outage, and zero otherwise. The speed, 24
channel model and the distance of the voice user to the serving sector shall also 25
be included in the curve. In case of SHO, the distance from the mobile to the 26
closest serving sector shall be used. 27
5. The outage probability for each user. Note that this value can be calculated 28
from the curve described in previous item. 29
The channel model and speed of a voice user are randomly chosen according to the pre-30
determined distributions specified in Table 2.2.2-1. 31
3.3.2.3.2 Mixed Voice and Data Services 32
In order to fully evaluate the performance of a proposal with mixed data and voice services, 33
simulations are repeated with different loads of voice users. The following outputs shall be 34
generated and included in the evaluation report. 35
1. The following cases shall be simulated: no voice users (i.e., data only), voice users 36
only (i.e., number of voice users equal to voice capacity), and 0.5Nmax or 0.8Nmax 37
voice users with data users, where Nmax is the voice capacity. 38
3GPP2 C.R1002-A v1.0
64
2. For each of the above case, all corresponding output matrices defined previously are 1
generated, whenever it is applicable. 2
In addition, the following output shall also be generated and included in the evaluation 3
report: 4
1. A curve of sector data throughput vs. the number of voice users is generated, where 5
the sector data throughput is defined as above. 6
3.3.2.4 Mixed Rev. 0 and Rev. A Mobiles (1xEV-DO Systems Only) 7
For proposals that support backward compatibility with 1xEV-DO (Rev. 0) ATs, the 8
following mixed Rev. 0 and Rev. A user tests shall be simulated using the standard system 9
layout and channel mix specified in section 2.2.2: 10
1. 8 full-buffer Rev. 0 ATs and 8 full-buffer Rev. A ATs per sector. 11
2. 16 Rev. 0 full-buffer ATs per sector. 12
3. 16 Rev. A full-buffer ATs per sector. 13
The applicable statistics required by the output matrices specified in Appendix M shall be 14
generated and presented for the purpose of evaluation, as well as the associated statistics 15
required for the evaluation report specified in 3.3.2.2. In addition, the following statistic 16
shall be reported: 17
i. Avg_TP_Rev0_Mix: Average throughput of Rev. 0 ATs in scenario 1. 18
ii. Avg_TP_RevA_Mix: Average throughput of Rev. A ATs in scenario 1. 19
iii. )_Re__
_Re__,
_0Re__
_0Re__(
OnlyvATPAvg
MixvATPAvg
OnlyvTPAvg
MixvTPAvg: 20
where Avg_TP_Rev0_Only denotes the average throughput per AT from scenario 2, 21
Avg_TP_Rev_A_Only denotes the average throughput per AT from scenario 3. 22
3.3.2.5 UMB Systems Only 23
The default bandwidth for the evaluation with full buffer traffic model is 5 MHz. The 24
simulation shall be performed with the following number of users: 25
c. Number of full buffer traffic users for 5 MHz – 1, 4, 8, 16, 32 user(s)/sector 26
d. Number of full buffer traffic users for 10 MHz – 1, 4, 8, 32 user(s)/sector 27
For the evaluation with mixed traffic model, the following user configuration shall be used: 28
c. 100 VoIP users/sector + 32 full buffer users/sector for 5MHz 29
d. 16 VT users/sector + 16 full buffer users/sector for 5MHz 30
If the proponent uses the transmit power headroom, the power headroom used should be 31
identified. 32
3GPP2 C.R1002-A v1.0
65
3.3.2.6 Link Level Output 1
All link level curves depicting the Eb/(No+Io) (from both receiving antennas of the sector) 2
vs. FER generated from reverse link level simulator are to be provided. 3
3.4 Calibration Requirements 4
Proponents and evaluators of any proposed RL design should follow the following 5
calibration steps to ensure a consistent evaluation process. 6
3.4.1 Link Level Calibration 7
All link-level data (in the form of Excel spreadsheet) should be submitted. One set of data 8
for use by all evaluators. 9
3.4.2 System Level Calibration 10
Each evaluator of the reverse link proposal(s) should submit the following calibration data 11
for the simulator. 12
3.4.2.1 1xEV-DV System Calibration 13
The capacity of the RL shall be simulated for voice-only users. The simulation environment 14
and assumptions shall be the same as those defined herein. Under these assumptions the 15
capacity of RL Nmax shall be computed and compared for voice-only users according to the 16
definition of outage in Appendix C and the rise-over-thermal criteria. Besides the capacity 17
number Nmax, the outages and the rise-over-thermals shall be included for various system 18
loads, from 5 mobiles per sector to Nmax + 3 MSs per sector with an increment of 5 mobiles 19
or less. 20
The following cases shall be simulated for a Rev. C system with F-FCH scheduling loaded 21
with data-only MSs. Scheduler operation is described in Appendix Q. 22
A single full buffer MS in each sector of the whole system in each of the fading 23
conditions at the following two locations, where the location of the MS relative to the 24
sector is the same across the sectors: 25
1. On the line between the sector and its closest neighboring cell at 0.3125 km from the 26
cell. (1/8 the way to the neighbor cell); 27
2. On the line between the sector and its closest neighboring cell at 1.25 km from the 28
cell (half way between these two cells). 29
Shadowing is zero dB for all path loss computations for this step. 30
Two full-buffer MSs positioned at the two fixed locations in the preceding step, respectively, 31
in each sector in the system. These MSs in the whole system for a simulation run shall 32
have the same fading model and the results for each of the fading models shall be 33
submitted. Shadowing is zero dB for all path loss computations for this step. 34
3.4.2.2 1xEV-DO System Calibration 35
The throughput performance for 1xEV-DO Rev. 0 reverse-link (RL) shall be simulated with 36
data-only users. The simulation shall be done for 4 and 16 full-buffer access terminals 37
3GPP2 C.R1002-A v1.0
66
(ATs) per sector16. The DRC gain shall be –1.5 dB for non-handoff ATs and –3 dB for ATs in 1
soft-handoff as mentioned in Table 3.1.2-1. For the purpose of calibration, the target rise-2
over-thermal (ROT) threshold that is used to set the RAB shall be 5 dB17. The received ROT 3
shall be updated per slot as the maximum per antenna ROT processed through a first-order 4
IIR filter with a time constant of 13.3333... ms (8 slots). The data-rate transition 5
probabilities in Table 3.4.2-1 shall be used with the default 1xEV-DO RL MAC algorithm: 6
Table 3.4.2-1 Default 1xEV-DO RL MAC Transition Probabilities 7
Transition009k6_019k2 0.5020
Transition019k2_038k4 0.2510
Transition038k4_076k8 0.1255
Transition076k8_153k6 0.0314
Transition019k2_009k6 0.0314
Transition038k4_019k2 0.0627
Transition076k8_038k4 0.1255
Transition153k6_076k8 1.0000
Unless otherwise specified, all channel structures and attributes shall follow the default 8
setting in 1xEV-DO Specification in document 3GPP2 C.S0024, Version 4.0. For the 9
specified mix of the five standard channel models in Section 2.2.2, the statistics required by 10
the output matrices specified in appendix M shall be generated and presented for the 11
purpose of evaluation, as well as the associated statistics required for the evaluation report 12
specified in Section 3.3.2.2. 13
3.4.2.3 UMB System Calibration 14
The following cases shall be simulated in the reverse link of UMB system. 15
a. Full buffer traffic only – 10AT, 10MHz 16
AWGN 17
Ped-B 3km/h and Veh-A 120km/h 18
b. VoIP traffic only – 100AT, 5MHz 19
Ped-B 3km/h and Veh-A 120km/h 20
Layout files for AN and AT are used[43]. The pathloss in the layout files contains 21
propagation loss, shadowing, and antenna pattern, but does not contain AN and AT 22
16 The choice of 16 ATs is used only for calibration purpose. It does not serve as an indication of the
maximum attainable 1xEV-DO RL capacity.
17 The purpose of this threshold is for calibration purpose only. This may or may not meet the
requirement of exceeding 7 dB ROT less than 1% of the time.
3GPP2 C.R1002-A v1.0
67
antenna gains as well as "other losses" (in total 4dB). Hence, 4 dB should be subtracted 1
from every path-loss value in the layout file. 2
A source file is used for VoIP traffic generation[42]. Maximum C/I of 28dB is used. 3
Reference link curve for FER prediction is used[44]. 8-tile curves are used for full buffer 4
simulation and 1-tile curves are used for VoIP simulation. 5
Details of parameters used in the calibration are summarized in [45]. 6
Output metrics used in the calibration follows the format of [46]. 7
3.4.2.3.1 Scheduler 8
A simplified scheduler is used in the calibration of UMB system simulators. The scheduler 9
is described for the following cases 10
a. Full buffer users only 11
b. VoIP users only 12
For the calibration purpose, closed loop precoding and/or subband scheduling is not 13
considered. LAB assignment errors are also ignored. Extended transmissions are not 14
supported, even for low geometry users. In the case of VoIP traffic, multiple tile 15
assignments are not supported. 16
The names of functions and variables are usually detailed enough to represent their 17
functionality. 18
Full buffer users only scheduler 19
Unlike the forward link full buffer case, in the reverse link full buffer case, due to reverse 20
link link budget, an AT may only be able to support transmitting in part of the bandwidth. 21
Therefore, the scheduler may need to schedule multiple ATs in the same interlace. Since 22
supplementary assignment is not supported, we only allow at most one node to be assigned 23
to an AT in each interlace. 24
For calibration, we use a fixed delta value feedback. We target -1dB CtoI. A packet format 2 25
will always be selected. 26
Following provides the pseudo-code for the scheduler used for RL full buffer users only case. 27
//Constants:
MaxNumLABs; //Maximum number of LABs allowed in a frame
N; //Number of ATs in the sector
numInterlace; //Number of interlaces (8)
//Scheduler function for full buffer users only
SchedulerFullBufferRL(){
//Check existing persistent assignment. If there is a decoding error, the assignment is assumed
//lost. Also check if the AT can support at least one tile assignment. If not, assume the
//persistent assignment lost.
UpdateExistingPersistentAssignments();
//Update channel desirability
3GPP2 C.R1002-A v1.0
68
for (i=1,…,N){
Priority[i] = RLCtoIMeasurement[i]
/ (AverageRLCtoIMeasurement[i] * ResourceAssignedToAT[i]);
//RLCtoIMeasurement shall be interpreted as spectral efficiency corresponds to the
//measured C/I. The measured C/I is computed from CDMA segment
//pCoT measurement, AT reported Delta_slow value, and IoT measurement.
//AverageRLCtoIMeasurement is an average of RLCtoIMeasurement with an
//one tap IIR filter with time constant 0.6 second.
//ResourceAssignedToAT is the actual time-frequency resource assigned to the AT
//counted from the beginning of the simulation.
}
//Compute maximum supportable assignment size
for (i=1,…,N){
maxAssignSize[i]=ComputeMaxAssignSize(PowerAmplifyerHeadroom[i])
}
//Count the number of LABs used, including RLABs (and possibly those used by VoIP users)
numLABs=NumUsedLABs();
//Find the list of ATs to be scheduled in the frame
toBeScheduled = FindATList(Priority[1,…,N]);
//Find the list of ATs that needs to be scheduled. This includes all ATs that
//are not in an active HARQ transmission. The output list is ordered in
//the decreasing order of Priority.
//Main loop of scheduling
while ( (toBeScheduled not empty) &&
(numLABs < MaxNumLABs) ){
idx=toBeScheduled.PopFirst(); //Consider the highest priority AT
if ( HasPersistentAssignmentInInterlace(idx) &&
(MaxNumLABs – numLABs < toBeScheduled.size())){
//The highest priority AT already has a persistent assignment, and there are too many
//ATs need to be scheduled, keep the persistent assignment.
continue;
}
//Find usable channel nodes for an AT. If the AT has persistent assignment in the interlace,
//assume the assignment is released
usableChannelNodeList1=FindUsableNodesAssumePersistent(idx);
if ( max(usableChannelNodeList1) < maxAssignSize[idx] ){
//The AT can support larger nodes than found, try find usable nodes again assuming
//all persistent assignments in the frame that are not in active transmission are released.
//Persistent assignments that are scheduled previously in this frame and the persistent
//assignments decided to keep in the frame are not released.
usableChannelNodeList2=FindUsableNodesAssumeNoPersistent();
//If the second round to find usable nodes can find something larger, use that list
if (max(usableChannelNodeList1)<max(usableChannelNodeList2)){
usableChannelNodesList1=usableChannelNodesList2;
3GPP2 C.R1002-A v1.0
69
}
}
//Pick in the list the largest node that does not exceed the maxAssignSize supportable.
//If the node is same size as the persistently assigned node (if available), use the same node.
node=PickBestNode( usableChannelNodeList1, maxAssignSize[idx]);
//Update the assignment to the AT. Fix packet format to 2, and
UpdateAssignment( idx,node,RLCtoIMeasurement[idx],
PowerAmplifierHeadroom[idx]){
//Check if the assignment is the same as the existing persistent one. If yes, do not send LAB.
if (AssignmentNotSameAsBefore(idx, node, packetFormat)){
SendLAB(idx, node, );
numLAB++;
}else{
//Same assignment as the persistent one. Keep the persistent assignment. Do nothing.
continue;
}
}
//All persistent assignment not affected by this function are still effective.
}
1
VoIP users only scheduler 2
For VoIP user, each AT is assigned one tile (base node) in one of the eight interlaces at a 3
time. 4
All VoIP assignments are persistent. When assigned, the tile-interlace pair is reserved for 5
the AT till it is explicitly de-assigned. A VoIP assignment will be explicitly de-assigned using 6
F-ACKCH. This is not modeled in the simulator. 7
The buffer size and time stamp of an AT is reported in band. The REQ is used only when 8
the buffer information is not included in the last traffic transmission. 9
When there is no more empty tile-interlace pair available, an assigned AT can be switched 10
out if there is no new packet arrives for the AT for the recent 25ms. 11
Power boost is not applied for HARQ re-transmissions for calibration purpose. 12
For VoIP traffic transmission, two packet formats are used and are optimized for EVRC 13
vocoder. Packet format 5 is used for full-rate and half-rate. Packet format 2 is used for 14
quarter-rate and 1/8-rate. 15
The scheduling algorithm is summarized in as follows: 16
Int MaxNumLABs; //Maximum number of LABs allowed in a frame
double frameInterval; //Time duration of a frame
int N; //Number of ATs in the sector
int numInterlace; //Number of interlaces (8)
//Scheduler function for VoIP users only. Old packet expiration shall be done in each AT
3GPP2 C.R1002-A v1.0
70
void SchedulerVoIPRL(){
//Check existing persistent assignment. If there is a decoding error, the assignment is assumed
//lost.
UpdateExistingPersistentAssignments();
//Update buffer and timer, using REQ and in-band signaling
UpdateBufferAndTimer(REQ[1,…,N]);
//Find available channel nodes
usableBaseNodeList = FindUsableNodes(timer);
//Find the usable channel nodes. Return a list of base nodes. In the list, the
//empty base nodes will be in the beginning. The releasable base nodes
//(persistently assigned to VoIP AT but not used in the frame and no REQ
//the AT in the recent 25ms). The releasable base nodes
//are listed in the order of last packet arrive time (earliest first).
//Find ATs to be scheduled in the frame
ToBeScheduled=FindATList(REQ[1,…,N]);
//Find the list of AT that needs to be scheduled. Generate a list of ATs
//that are not assigned. The output is ordered in decreasing order of the time
//stamp of the first packet in the buffer.
//Main loop to schedule VoIP ATs
int numLABs = LABinFL();
while ( (usableBaseNodeList not empty) &&
(numLABs < MaxNumLABs) &&
(ToBeScheduled not empty)){
//Pick the first AT in the ToBeScheduled list and remove it from the list
int idx=ToBeScheduled.PopFirst();
//Pick the first base node in the usableBaseNodeList
node=usableBaseNodeList.PopFirst();
numLABs++;
powerDensity=ComputePowerDensity();
//Compute the power density to achieve 1.5dB per antenna CtoI
//assuming the packet format for full rate packet;
SendLAB(idx, node, powerDensity);
}
//All persistent assignment not affected by this function are still effective.
}
1
3.5 1xEV-DO Baseline Simulation Procedures 2
3.5.1 Access Terminal Requirements and Procedures: 3
1xEV-DO Revision 0 access terminal (AT) shall follow procedures specified in 1xEV-DO 4
Specification in document 3GPP2 C.S0024, version 4.0 unless otherwise specified below: 5
3GPP2 C.R1002-A v1.0
71
1. Default setting of DRC channel gain: The DRC channel shall be modeled as a 1
continuous channel with a gain of –1.5 dB w.r.t. pilot for terminals that are not in 2
soft-handoff (DRC Length = 2 slots) and –3 dB w.r.t. pilot for terminals that are in 3
soft-handoff (DRC Length = 4 slots). 4
2. Rate transition rules: The data rate of each AT shall be adjusted according to the 5
default 1xEV-DO RL MAC rules and the received RAB information using the rate 6
transition probabilities specified in Table 3.4.2-1. 7
3. The effect of RL ACK channel shall be ignored due to its negligible impact on RL 8
capacity. 9
4. For baseline purposes, the procedure for determining whether a specific data-rate is 10
power-limited for the AT shall be carried out using the default MAC algorithm by 11
assuming a fixed transmit power headroom of 3 dB. The estimate for pilot transmit 12
power used in this process shall be the received pilot power averaged over the most 13
recent frame (16 slots). 14
3.5.2 Access Network Requirements and Procedures: 15
1xEV-DO Revision 0 access network (AN) shall follow procedures specified in 1xEV-DO 16
Specification in document 3GPP2 C.S0024, version 4.0 unless otherwise specified below: 17
1. The target rise-over-thermal (ROT) threshold that is used to set the RAB bit shall be 18
set to meet the criterion that the percentage of time the received ROT above 7 dB 19
does not exceed 1% over the entire simulation run for simulation involving 16 full-20
buffer users per sector. This specific ROT threshold shall be used for all simulation 21
scenarios specified in 3.5.3. 22
2. The received ROT shall be measured and updated per slot as the maximum per 23
antenna ROT processed through a first-order IIR filter with a time constant of 24
13.3333... ms (8 slots). 25
3. The base-station compares the measured ROT (at the output of the IIR filter) with 26
the target threshold every frame (16 slots). If the ROT exceeds the threshold, the 27
RAB bit status shall be set to busy, and otherwise it shall be set to not busy. The 28
updated RAB is transmitted over one frame (16 slots) to all ATs within the sector. 29
4. The impact of imperfect RAB channel shall be modeled in the network simulation 30
using a RAB decoder with dynamic FL modeling. The dynamic modeling shall 31
assume the same multipath and Doppler channel model on RL and FL. The fading 32
process on FL and RL shall be assumed as independent. 33
3.5.3 Simulation Procedures 34
The simulation shall be carried out using the following settings based on the layout 35
configuration and method specified in section 3.318: 36
1. 16 full-buffer ATs per sector. 37
18 Settings 3, 4, and 5 apply standard traffic mixes specified in 4.2.1.2.
3GPP2 C.R1002-A v1.0
72
2. 4 full-buffer ATs per sector. 1
3. 4 HTTP users, 10 WAP users, 3 MNG users, and FTP users arrive in a Poisson 2
manner described in section 4.2. 3
4. 8 HTTP users, 20 WAP users, 6 MNG users, and FTP users arrive in a Poisson 4
manner describe in section 4.2. 5
5. No HTTP, WAP or MNG users, and FTP users arrive in a Poisson manner described 6
in section 4.2. 7
For each setting, the statistics according to the output matrices in appendix M shall be 8
generated and presented as well as the associated statistics required for the evaluation 9
report specified in section 3.3.2. The simulation time for each run in scenarios 3, 4, and 5 10
shall be 100 seconds (60000 slots). FTP user arrival rate shall be determined in the same 11
way as that specified in section 3.3.1.2. In addition, the link-level PER vs. Eb/Nt curves 12
used in the network simulator shall be presented. The link-level curve shall be run with a 13
target FER of 5% or less. 14
. 15
3GPP2 C.R1002-A v1.0
74
4 TRAFFIC SERVICE MODELS 1
4.1 Forward Link Services 2
4.1.1 Service Mix (1xEV-DV Systems Only) 3
A configurable fixed number of voice calls are maintained during each simulation run. Data 4
sector throughput is evaluated as a function of the number of voice users supported. The 5
following cases shall be simulated: no voice users (i.e., data only), voice users only (i.e., the 6
number of voice users equals to voice capacity), and average 0.5Nmax or 0.8Nmax voice 7
users per sector plus data users, where Nmax is the voice capacity defined in Appendix C. 8
The data users in each sector shall be assigned one of the four traffic models: WAP 9
(56.43%), HTTP (24.43%), FTP (9.29%), near real time video (9.85%), with the respective 10
probabilities in parentheses. 11
4.1.2 TCP Model 12
Since FTP and HTTP use TCP as their transport protocol, a TCP traffic model is introduced 13
to more accurately represent the distribution of TCP packets for the FTP and HTTP traffic 14
models described in the next sections. 15
The TCP connection set-up and release protocols use a three-way handshake mechanism 16
as described in Figure 4.1.2-1 [16]. The amount of outstanding data that can be sent 17
without receiving an acknowledgement (ACK) is determined by the minimum of the 18
congestion window size and the receiver window size. After the connection establishment is 19
complete, the transfer of data starts in slow-start mode with an initial congestion window 20
size of 1 segment. The congestion window increases by one segment for each ACK packet 21
received by the sender regardless of whether the packet is correctly received or not, and 22
regardless of whether the packet is out of order or not. This results in an exponential 23
growth of the congestion window. This process is illustrated in Figure 4.1.2-2. 24
3GPP2 C.R1002-A v1.0
75
Server
(Host/BS)
Client
(MS)
SYN(K)
SYN(J)
ACK(K+1)
ACK(J+1)+
HTTP GET
FIN(M)
ACK(M+1) FIN(N)
ACK(N+1)
J
K+1
ACK PSH RST SYN FINURG
20 bytes
[SYN(J), ACK(K+1)]
J
K+1
ACK PSH RST SYN FINURG
20 bytes
J
K+1
ACK PSH RST SYN FINURG
20 bytes
[SYN(J), ACK(K+1)]
HTTP Response
1
Figure 4.1.2-1 Control Segments in TCP Connection Set-up and Release 2
3GPP2 C.R1002-A v1.0
76
l
l
l
rt
rt
rt
l
Client Router Server
l
Access Link Connecting Link
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
1
23
45
67
89
1011
1213
1415
1617
1819
2021
2223
1
Figure 4.1.2-2 TCP Flow Control During Slow-Start; l = Transmission Time over the 2
Access Link; rt = Roundtrip Time 3
3GPP2 C.R1002-A v1.0
77
The round-trip time in Figure 4.1.2-2, rt, consists of two components: 1
rt = c + l (4.1.2-1) 2
where c = the sum of the time taken by an ACK packet to travel from the client to the 3
server and the time taken by a TCP data segment to travel from the server to the base 4
station router; l = the transmission time of a TCP data segment over the access link from 5
the base station router to the client. 6
In this model of TCP, c is modeled as an exponentially distributed random variable with a 7
mean of 50 ms; l is determined by the available access link throughput. Also, it should be 8
mentioned that the detailed specifics of congestion control and avoidance has not been 9
modeled; only the slow-start part has been modeled. It is also assumed that, the receiver 10
window size is very large and hence is not a limitation. 11
From Figure 4.1.2-2, it can be observed that, during the slow-start process, for every ACK 12
packet received by the sender two data segments are generated and sent back to back. 13
Thus, at the base station, after a packet is successfully transmitted, two segments arrive 14
back-to-back after an interval c. Based on this observation, the packet arrival process at 15
the base station for the download of an object is shown in Figure 4.1.2-3. It is described as 16
follows: 17
1. Let S = size of the object in bytes. Compute the number of packets in the 18
object, N = S/(MTU-40). Let W = size of the initial congestion window of 19
TCP1. 20
1 Compressed TCP header (5 bytes from 40 bytes) and PPP framing overhead (7 bytes + ceiling
(size(payload + compressed TCP header + 6)/128)) shall be transmitted over the air in addition to the
payload (MTU – 40) bytes.
For the 1500 MTU size:
1460 Payload (MTU – 40)
+ 5 Compressed TCP header
+ 7 PPP headers and CRC
+ 12 PPP escaping = ceiling ( (1460 + 5 + 6)/128)
1484 bytes above the multiplex sublayer
For the 576 MTU size:
536 Payload (MTU – 40)
+ 5 Compressed TCP header
+ 7 PPP headers and CRC
+ 5 PPP escaping = ceiling ( (536 + 5 + 6)/128)
553 bytes above the multiplex sublayer
3GPP2 C.R1002-A v1.0
78
2. If N>W, then W packets are put into the queue for transmission; otherwise, 1
all packets of the object are put into the queue for transmission in FIFO 2
order. Let P=the number of packets remaining to be transmitted. If P=0, go to 3
step 6. 4
3. Wait until a packet of the object in the queue is transmitted over the access 5
link. 6
4. Schedule arrival of next two packets (or the last packet if P=1) of the object 7
after an interval of c. If P=1, then P=0, else P=P-2. 8
5. If P>0 go to step 3. 9
6. Preserve PW = N+W, as the size of the congestion window to be used by 10
persistent TCP connections. 11
7. Return. 12
3GPP2 C.R1002-A v1.0
79
S = size of object in
bytes
N = CEIL(S/(MTU-40));
W = initial TCP window
size;
C = min(N,W);
Enqueue C packets for
transmission;
P=N-C;
yes
no
t = current time;
Schedule arrival of
min(P,2) next packets
of the object at the
queue at T=t+Tc;
P = P - min(P,2)
P > 0 ?
Wait until a packet of
the object in the queue
is transmitted
PW = N+W
return PW
PW is the size of the
TCP window returned for
use in HTTP/1.1-
persistent connection
1
Figure 4.1.2-3 Packet Arrival Process at the Base Station for the Download of an 2
Object Using TCP; PW = the Size of the TCP Congestion Window at the End of 3
Transfer of the Object; Tc=c (Described in Figure 4.1.2-2) 4
3GPP2 C.R1002-A v1.0
80
4.1.3 HTTP Model 1
4.1.3.1 HTTP Traffic Model Characteristics 2
A sessionFirst packet of the
sessionLast packet of the
session
Instances of packet
arrival at base station
A packet callreading time
3
Figure 4.1.3-1 Packet Trace of a Typical Web Browsing Session 4
Figure 4.1.3-1 shows the packet trace of a typical web browsing session. The session is 5
divided into ON/OFF periods representing web-page downloads and the intermediate 6
reading times. In Figure 4.1.3-1, the web-page downloads are referred to as packet calls 7
and are denoted as such in [18]. These ON and OFF periods are a result of human 8
interaction where the packet call represents a user‘s request for information and the 9
reading time identifies the time required to digest the web-page. 10
As is well known, web-browsing traffic is self-similar. In other words, the traffic exhibits 11
similar statistics on different timescales. Therefore, a packet call, like a packet session, is 12
divided into ON/OFF periods as in Figure 4.1.3-2. Unlike a packet session, the ON/OFF 13
periods within a packet call are attributed to machine interaction rather than human 14
interaction. As an example, consider a typical web-page from the Wall Street Journal (WSJ) 15
Interactive edition depicted in Figure 4.1.3-3. This web-page is constructed from many 16
individually referenced objects. A web-browser will begin serving a user‘s request by 17
fetching the initial HTML page using an HTTP GET request. After receiving the page, the 18
web-browser will parse the HTML page for additional references to embedded image files 19
such as the graphics on the tops and sides of the page as well as the stylized buttons. The 20
retrieval of the initial page and each of the constituent objects is represented by ON period 21
within the packet call while the parsing time and protocol overhead are represented by the 22
OFF periods within a packet call. For simplicity, the term ―page‖ will be used in this paper 23
to refer to each packet call ON period. As a rule-of-thumb, a page represents an individual 24
HTTP request explicitly initiated by the user. The initial HTML page is referred to as the 25
―main object‖ and the each of the constituent objects referenced from the main object are 26
referred to as an ―embedded object‖. 27
3GPP2 C.R1002-A v1.0
81
Dpc
Nd
packet callpacket call
embedded objects
(Reading Time)
main object
1
Figure 4.1.3-2 Contents in a Packet Call 2
3
4
Figure 4.1.3-3 A Typical Web Page and Its Content 5
The parameters for the web browsing traffic are as follows: 6
SM: Size of the main object in a page 7
SE: Size of an embedded object in a page 8
Nd: Number of embedded objects in a page 9
Dpc: Reading time 10
Tp: Parsing time for the main page 11
The packet traffic characteristics within a packet call will depend on the version of HTTP 12
used by the web servers and browsers. Currently two versions of the protocol, HTTP/1.0 13
and HTTP/1.1[7,8,11,12,13], are widely used by the servers and browsers. These two 14
3GPP2 C.R1002-A v1.0
82
versions differ in how the transport layer TCP connections are used for the transfer of the 1
main and the embedded objects as described below. 2
In HTTP/1.0, a distinct TCP connection is used for each of the main and embedded objects 3
downloaded in a web page. Most of the popular browser clients download the embedded 4
objects using multiple simultaneous TCP connections; this is known as HTTP/1.0-burst 5
mode transfer. The maximum number of such simultaneous TCP connections, N, is 6
configurable; most browsers use a maximum of 4 simultaneous TCP connections. If there 7
are more than N embedded objects, a new TCP connection is initiated when an existing 8
connection is closed. The effects of slow-start and congestion control overhead of TCP occur 9
on a per object basis. 10
In HTTP/1.1, persistent TCP connections are used to download the objects, which are 11
located at the same server and the objects are transferred serially over a single TCP 12
connection; this is known as HTTP/1.1-persistent mode transfer. The TCP overhead of slow-13
start and congestion control occur only once per persistent connection. 14
4.1.3.2 HTTP Traffic Model Parameters 15
The distributions of the parameters for the web browsing traffic model were determined 16
based on the survey of the literature on web browsing traffic characteristics [12,13]. These 17
parameters are described in Table 4.1.3-11 18
1 Truncated lognormal distribution for the main and embedded object size parameters shall be used.
The truncation points are as follows:
For main object: Maximum = 2,000,000 bytes, minimum = 100 bytes
For embedded objects: Maximum = 2,000,000 bytes, minimum = 50 bytes
3GPP2 C.R1002-A v1.0
83
Table 4.1.3-1 HTTP Traffic Model Parameters 1
Component Distribution Parameters PDF
Main object
size (SM)
Truncated
Lognormal
Mean = 10710 bytes
Std. dev. = 25032
bytes
Minimum = 100 bytes
Maximum = 2 Mbytes
Embedded
object size
(SE)
Truncated
Lognormal
Mean = 7758 bytes
Std. dev. = 126168
bytes
Minimum = 50 bytes
Maximum = 2 Mbytes
Number of
embedded
objects per
page (Nd)
Truncated
Pareto
Mean = 5.64
Max. = 53
Note: Subtract k from the
generated random value to
obtain Nd
Reading time
(Dpc)
Exponential Mean = 30 sec
Parsing time
(Tp)
Exponential Mean = 0.13 sec
Note: When generating a random sample from a truncated distribution, discard the random 2
sample when it is outside the valid interval and regenerate another random sample. 3
From the literature [13], it was found that, most of the web pages are downloaded in 4
HTTP/1.0-burst mode or HTTP/1.1-persistent mode. From the statistics presented in these 5
literatures, it was concluded that, a 50%-50% distribution of the web pages between HTTP 6
1.0-burst mode and HTTP 1.1-serial mode will closely approximate the web browsing traffic 7
behavior in the Internet in the short term (in the time-frame of 1xEV-DV and 1xEV-DO 8
deployments). Also, based on some of the studies on packet size properties in the Internet 9
[10,19], it was observed that, the MTU sizes most prominently used by the TCP connections 10
35.8,37.1
0,22
ln2
exp2
1
x
x
xxf
17.6,36.2
0,22
ln2
exp2
1
x
x
xxf
55,2,1.1
,
,1
mk
mx
m
k
f x
mxk
x
kf x
033.0
0,
xex
f x
69.7
0,
xex
f x
3GPP2 C.R1002-A v1.0
84
in the Internet are 576 bytes and 1500 bytes. A distribution of 24%-76% of all web pages 1
for transfer using an MTU of 576 bytes and 1500 bytes1 along with the TCP control 2
segments of 40 bytes (as described in Figure 4.1.2-1) will closely approximate the packet 3
size characteristics for the downlink. Thus, the web traffic generation process can be 4
described as in Figure 4.1.3-4. 5
HTTP version ?
Download the main and
the embedded objects
using HTTP/1.0-burst
transport
Download the main
and the embedded
objects using HTTP/
1.1-persistent transport
HTTP/1.1HTTP/1.0
Create an HTML
page using the HTML
page statistics
50% 50%
MTU = 576 bytesMTU = 1500 bytes
MTU ?76% 24%
6
Figure 4.1.3-4 Modeling a Web Page Download 7
In the rest of this section, the packet arrival models are described which approximate the 8
packet arrival process on the downlink in HTTP/1.0-burst mode and HTTP/1.1-persistent 9
mode. 10
4.1.3.2.1 Packet Arrival Model for HTTP/1.0-Burst Mode 11
The download of the web page is modeled as follows: 12
1. The main object is downloaded with a new TCP connection whose initial 13
congestion window size is 1. 14
2. The embedded objects are partitioned into groups of four objects; each group is 15
called a composite object. 16
1 The MTU size remains fixed for a packet call or page. In other words, all objects in a page (both the main object and the embedded objects) are transferred using the same MTU size.
3GPP2 C.R1002-A v1.0
85
3. A composite object, made of b (1<=b<=4) embedded objects, is transferred using a 1
new TCP connection whose initial window size is b. 2
4. The transfer of the 1st composite object is initiated Tp+c second after the 3
transmission of the main object is completed. The transfer of subsequent composite 4
object begins c second after the transmission of the previous composite object is 5
completed. 6
The transfer of the main object, or of a composite object made of b embedded objects, is 7
modeled as shown in Figure 4.1.3-5. The process is described as follows: 8
1. Let S = size of the object in bytes. If it is the main object, set the initial TCP window 9
size m=1; if it is a composite object, set m=b, where b=the number of embedded 10
objects in the composite object. 11
2. Put m 40-byte SYN+ACK packets for the connection establishment into the queue 12
for transmission and wait until it is transmitted. 13
3. Transmit another set of m 40-byte packets [9]. 14
4. Begin transfer of the object (using the flowchart of Figure 4.1.2-3) c second later 15
with an initial TCP window size m. 16
5. When transfer of the object is completed, transmit m 40-byte FIN segment to initiate 17
closing of connection. 18
6. Wait for c second. 19
7. Transmit m 40-byte ACK packets to complete connection close. 20
3GPP2 C.R1002-A v1.0
86
S=size of the object in bytes;
m=1,for main object
=b, for a composite object made
of b embedded objects
Complete transmission of m 40-byte
SYN+ACK segments
Wait Tc
Complete transfer of object
content of size S bytes and
initial TCP window size W=m
(using Figure 4.1.2-3)
Complete
transmission of m
40-byte FIN
segment
Wait Tc
Complete
transmission of m
40-byte ACK
packets
end
Complete transmission of m 40-byte
control segments
1
2
Figure 4.1.3-5 Download of an Object in HTTP/1.0-Burst Mode 3
4.1.3.2.2 Packet Arrival Model for HTTP/1.1-Persistent Mode 4
The download of the web page is modeled as follows: 5
1. The main object is transferred using a new TCP connection whose initial window 6
size is 1; this TCP connection and its congestion window size is preserved. 7
2. All of the embedded objects are transferred, serially one after another, using the 8
preserved TCP connection. 9
3GPP2 C.R1002-A v1.0
87
3. The transfer of embedded objects begins Tp+c second after the completion of the 1
transfer of the main object. 2
4. The TCP congestion window size is preserved at the end of transmission of each 3
embedded object. 4
5. Transfer of next embedded object begins c second after the transmission of the 5
previous embedded object is completed. 6
6. At the end of transfer of all embedded objects, the client initiates release of the TCP 7
connection. 8
The transfer of the main and embedded objects are shown in Figure 4.1.3-6. It is described as 9
follows: 10
1. Let S = size of the main object in bytes. 11
2. Transmit a 40-byte SYN+ACK packet for the connection establishment. 12
3. Transmit another 40-byte packet. 13
4. Begin transfer of the main object (using the flowchart of Figure 4.1.2-3) c second 14
later with an initial TCP window size 1. 15
5. At the end of transfer of the main object, preserve the congestion window size PW; 16
6. Wait for Tp sec (the parsing time). 17
7. If Nd = 0 (there are no embedded objects), go to step 16. 18
8. Let I=1. 19
9. Transmit four 40-byte packets. 20
10. If I>Nd (the number of embedded objects) go to step 16 21
11. Wait for c second (for initiation of transfer of embedded object). 22
12. S = size of the I-th embedded object in bytes. 23
13. Begin transfer of the I-th embedded object (using the flowchart in Figure 4.1.2-3) 24
with initial TCP congestion window size of PW (the preserved congestion window). 25
14. At the completion of transfer of the I-th object, preserve the congestion window size 26
PW. 27
15. I = I+1; go to step 10 to initiate transfer of next embedded object. 28
16. Wait for c (delay for client initiated TCP close). 29
17. Transmit a 40-byte ACK packet for connection close. 30
18. Transmit a 40-byte FIN packet for connection close. 31
3GPP2 C.R1002-A v1.0
88
S=size of the i-th object in
bytes
Complete transfer of the i-th object of size S bytes with initial
TCP window size W=m (using the flowchart of Figure 4.1.2-3)
Complete transmission of a 40-byte control segment
Wait Tc
Complete transmission of a 40-byte control packets
end
i=1;
i < Nd ? i=i+1
m = PW (returned by TCP transfer process)
Transmit a 40-byte SYN+ACK segment
Transmit another 40-byte segment
Complete transfer of the main object of size S bytes with initial
TCP window size W=1(using the flow-chart of Figure 4.1.2-3);
Transmit four 40-byte packets;
Wat Tc
Wait Tc
Wait Tp
Nd>0 ?
yes
yes
no
no
m = PW (returned by TCP transfer process);
1
Figure 4.1.3-6 Download of Objects in HTTP/1.1-Persistent Mode 2
3GPP2 C.R1002-A v1.0
89
4.1.4 FTP Model 1
4.1.4.1 FTP Traffic Model Characteristics 2
In FTP applications, a session consists of a sequence of file transfers, separated by reading 3
times. The two main parameters of an FTP session are: 4
1. S : the size of a file to be transferred 5
2. Dpc: reading time, i.e., the time interval between end of download of the previous file 6
and the user request for the next file. 7
The underlying transport protocol for FTP is TCP. The model of TCP connection described 8
in section 4.1.2 will be used to model the FTP traffic. The packet trace of an FTP session is 9
shown in Figure 4.1.4-1. 10
Packet calls
Dpc
Packets of file 1 Packets of file 2 Packets of file 3
11
Figure 4.1.4-1 Packet Trace in a Typical FTP Session 12
4.1.4.2 FTP Traffic Model Parameters 13
The parameters for the FTP application sessions are described in Table 4.1.4-1. 14
3GPP2 C.R1002-A v1.0
90
Table 4.1.4-1 FTP Traffic Model Parameters 1
Component Distribution
Parameters
File size (S) Truncated Lognormal
Mean = 2Mbytes
Std. Dev. = 0.722 Mbytes
Maximum = 5 Mbytes
Reading time (Dpc)
Exponential Mean = 180 sec.
2
Based on the results on packet size distribution [10,19] (described in section 4.1.3.2), 76% 3
of the files are transferred using and MTU of 1500 bytes and 24% of the files are 4
transferred using an MTU of 576 bytes. For each file transfer a new TCP connection is used 5
whose initial congestion window size is 1 segment (i.e. MTU). The packet arrival process at 6
the base station is described by the flowchart in Figure 4.1.2-3 (in which the object 7
represents the file being transferred). The process for generation of FTP traffic is described 8
in Figure 4.1.4-2. 9
45.14,35.0
0,22
ln2
exp2
1
x
x
xxf
006.0
0,
xex
f x
3GPP2 C.R1002-A v1.0
91
Create a file using the
file size statistics in
Table 4.1.4-1
MTU ?
MTU = 1500 bytes MTU = 576 bytes
Complete transfer of the file
using a new TCP connection
with initial window size
W=1(as shown in Figure
4.1.2-3)
Wait Dpc
1
Figure 4.1.4-2 Model for FTP Traffic 2
4.1.5 WAP Model 3
Each WAP request from the browser is modeled as having a fixed size and causes the WAP 4
server to send back a response with an exponentially distributed response time. The WAP 5
gateway response time is the time between when the last octet of the request is sent and 6
when the first octet of the response is received from the WAP server. The response itself is 7
composed of a geometrically distributed number of objects, and the inter-arrival time 8
between these objects is exponentially distributed. Once the last object is received, the 9
exponentially distributed reading time starts, and it ends when the WAP browser generates 10
the next request. Figure 4.1.5-1 illustrates the data flow for the WAP traffic model and 11
Table 4.1.5-1 describes the distribution of the model parameters. 12
3GPP2 C.R1002-A v1.0
92
MS BS
PDSN/WAP
GatewayContent
Server
WAP Gateway
Response Time
(exponential)
Reading Time -
includes browser rendering delay
(exponential)
Request
Response, 1st object
(size is truncated pareto distributed)
HTTP Response
HTTP GET
N = #
objects
per
response
(geometr
ic)
Response, 2nd object
(size is truncated pareto distributed)
object inter-arriv al
time (exponential)
Response, Nth object
(size is truncated pareto distributed)
....
RequestHTTP GET
1
Figure 4.1.5-1 Packet Trace for the WAP Traffic Model 2
During the simulation period, the model assumes that each WAP user is continuously 3
active, i.e., making WAP requests, waiting for the response, waiting the reading time, and 4
then making the next request. 5
Table 4.1.5-1 WAP Traffic Model Parameters 6
Packet based information
types
Size of WAP request
Object size # of objects per response
Inter-arrival time between
objects
WAP gateway response time
Reading time
Distribution Deterministic Truncated Pareto
(Mean= 256 bytes, Max= 1400
bytes)
Geometric
P(k) = 0.5k
,
k>=1
Exponential Exponential Exponential
Distribution Parameters
76 octets K = 71.7 bytes, = 1.1
Mean = 2 Mean = 1.6 s Mean = 2.5 s Mean = 5.5 s
4.1.6 Near Real Time Video Model 7
The following section describes a model for streaming video traffic on the forward link. 8
3GPP2 C.R1002-A v1.0
93
Figure 4.1.6-1 describes the steady state of video streaming traffic from the network as 1
seen by the base station. Latency of starting up the call is not considered in this steady 2
state model. 3
T 2T (K-1)T0 KTT
B (Buffering Window)
Video Streaming Session (= simulation time)
DC (Packet
Coding Delay)
Packet Size
time
4
5
Figure 4.1.6-1 Video Streaming Traffic Model 6
A video streaming session is defined as the entire video streaming call time, which is equal 7
to the simulation time for this model. 8
Each frame of video data arrives at a regular interval T determined by the number of frames 9
per second (fps). Each frame is decomposed into a fixed number of slices, each transmitted 10
as a single packet. The size of these packets/slices is distributed as a truncated Pareto. 11
Encoding delay, Dc, at the video encoder introduces delay intervals between the packets of 12
a frame. These intervals are modeled by a truncated Pareto distribution. 13
The parameter TB is the length (in seconds) of the de-jitter buffer window in the mobile 14
station used to guarantee a continuous display of video streaming data. This parameter is 15
not relevant for generating the traffic distribution but is useful for identifying periods when 16
the real-time constraint of this service is not met. At the beginning of the simulation, it is 17
assumed that the mobile station de-jitter buffer is full with (TB x source video data rate) bits 18
of data. Over the simulation time, data is ―leaked‖ out of this buffer at the source video 19
data rate and ―filled‖ as forward link traffic reaches the mobile station. As a performance 20
criterion, the mobile station can record the length of time, if any, during which the de-jitter 21
buffer runs dry. The de-jitter buffer window for the video streaming service is 5 seconds. 22
Using a source video rate of 32 kbps, the video traffic model parameters are defined in 23
Table 4.1.6-1. 24
3GPP2 C.R1002-A v1.0
94
Table 4.1.6-1 Video Streaming Traffic Model Parameters 1
Information types
Inter-arrival time between the beginning of each frame
Number of packets
(slices) in a frame
Packet (slice) size
Inter-arrival time between
packets (slices) in a frame
Distribution Deterministic
(Based on
10fps)
Deterministic Truncated Pareto
(Mean= 50bytes, Max= 125bytes)
Truncated Pareto
(Mean= 6ms, Max= 12.5ms)
Distribution Parameters
100ms 8 K = 20bytes = 1.2
K = 2.5ms = 1.2
4.1.7 Voice Model (1xEV-DV Systems Only) 2
Voice follows a standard Markov source model. The voice activity factor is 0.403 with 29% 3
full rate, 60% eighth rate, 4% half rate, and 7% quarter rate. The corresponding transition 4
probabilities are defined in C.S0025 (TIA/EIA/IS-871). 5
Voice capacity shall be obtained based on three outage criteria (Short Term FER 15%, Per 6
User Outage 1%, and System Outage 3%) defined in Appendix C. 7
Ignoring other effects, energy consumption per bit for the four data rates in the voice 8
service is the same. That implies, for example, the required traffic Ec for the 4800 bps is 3 9
dB less than that required for 9600 bps. 10
Voice is modeled as C.S0002 (IS-2000) RC3 and RC4, power controlled, with and without 11
STS. Four possible combinations of RC3 and RC4, non-transmit-diversity, and STS shall 12
be simulated independently. Power control subchannel power consumption shall be 13
specified. The forward power control subchannel is sent at the same power level as the full 14
rate (9.6 kbps) frame when the MS is not in soft handoff. That subchannel is sent at a level 15
3 and 5 dB higher than the 9.6 kbps level when in two-way and three-way soft handoff, 16
respectively. The reverse link power control subchannel is sent on the Reverse Pilot 17
Channel in a TDM fashion with the same power level as the pilot. Any other arrangements 18
in the proposals shall be properly accounted for with clear definitions for power 19
consumption. The power consumption by all reverse link overhead channels, including the 20
Reverse Pilot Channel shall be specified and justified. 21
4.1.8 Delay Criteria 22
Except for WAP users, all 1xEV-DV and 1xEV-DO packet data (HTTP, FTP, or near real time 23
video) users shall satisfy the following delay criterion: no more than 2% of the users shall 24
get less than 9600 bps throughput (goodput). The throughput will be the user's packet call 25
throughput, except in the case where there is no arrival process (FTP users are persistent) 26
in which case it will be the throughput averaged over the simulation time. 27
Neal real time video users shall also satisfy the performance criteria defined in section 28
4.1.8.1. 29
WAP users shall satisfy the delay criterion defined in section 4.1.8.2. 30
3GPP2 C.R1002-A v1.0
95
4.1.8.1 Performance Criteria for Near Real Time Video 1
Video playout buffers introduce a delay between receipt of frames and the frame playout. 2
This absorbs variations in the data arrival pattern and permits a continuous playout of the 3
frames. The actual design of these playout buffers involves a number of factors (including 4
reset policies when the buffer runs dry) and is specific to the mobile. To avoid modeling 5
such implementation details, we focus on what the BS scheduler must do to generally 6
accommodate this continuous playout. Therefore, the scheduler should transmit an entire 7
video frame within 5 second of receipt of the entire frame (i.e., receipt of the last octet of the 8
last slice of the frame). If a frame exceeds the 5-second requirement, the scheduler 9
discards the remainder of the frame that has not yet been transmitted. The size and arrival 10
statistics for the video frames are defined in section 4.1.6. 11
Therefore, the performance requirement is that the fraction of video frames that are not 12
completely transmitted within 5 seconds of their arrival at the scheduler shall be less than 13
2% for each user. All users shall meet the above performance requirement. 14
4.1.8.2 Delay Criterion for WAP Users 15
No more than 2% of the users shall get less than 4800 bps throughput (goodput). The 16
throughput will be the user's packet throughput. The packet throughput of a user is 17
defined as the ratio of the total number of information bits that an user successfully 18
receives and the accumulated delay for all packets for the user, where the delay for an 19
individual packet is defined as the time between when the packet enters the queue at 20
transmitter and the time when the packet is received successively by the mobile station. If a 21
packet is not successfully delivered by the end of a run, its ending time is the end of the 22
run. Using the terminology defined in Appendix D, the packet throughput for user(m, n) 23
can be obtained as 24
Packet throughput for user(m,n)
),(
1
),,(
1
),(
1
),,(
1
),,,(),,,(
),,,(
nmK
k
knmL
l
nmK
k
knmL
l
lknmTAlknmTD
lknmB
(4.1.8-1) 25
4.2 Reverse Link Services 26
4.2.1 Service Mix (1xEV-DV Systems Only) 27
A configurable fixed number of voice calls are maintained during each simulation run. 28
Data sector throughput is evaluated as a function of the number of voice users supported. 29
The following cases shall be simulated: 30
No voice users (i.e., data only) 31
Voice users only 32
0.5Nmax or 0.8Nmax voice users per sector plus data users. 33
3GPP2 C.R1002-A v1.0
96
4.2.1.1 Data Model 1
The throughput contribution of an individual MSi is: 2
ii
i
Correctly Re ceived Bits from MSAve Throughput for MS
Simulation Time of MS (4.2.1.1-1) 3
where 4
The simulation time of a WAP, HTTP, or MNG MS is the entire duration of the 5
simulation run. 6
The simulation time of an FTP upload mobile station is the duration that mobile 7
station is on the system, excluding the initial setup period. 8
4.2.1.2 Traffic Model 9
Data traffic is divided into two categories: 10
Forward link supporting traffic: HTTP/WAP requests and TCP ACKs1. Among these 11
users, 70% corresponds to WAP users and 30% corresponds to HTTP users. The 12
traffic configurations to be simulated are given in Table 4.2.1-1, where Nmax is the 13
voice capacity per sector, as defined in Appendix C2. 14
Table 4.2.1-1: Traffic Configurations 15
HTTP WAP MNG Voice3 FTP
0 0 0 0 Poisson arrival
4 10 3 0 Poisson arrival
8 20 6 0 Poisson arrival
4 10 3 max5.0 N Poisson arrival
2 5 0 max8.0 N Poisson arrival
16
Reverse link specific traffic: FTP upload users. FTP Upload users represent mobiles 17
uploading a file with FTP, or sending an email attachment. The arrival of FTP upload 18
users is modeled as a Poisson arrival process with arrival rate . The Poisson arrival 19
1 The maximum number of users in this class is obtained according to the FL traffic mix that needs
support on the RL. The maximum number of users on the FL is assumed to be 42 users
(approximately the FL capacity for data-only users, mix of channel models and mix of traffic models).
Since WAP+HTTP compose 80.86% of the FL traffic, the maximum number of users to support on RL
is 34.
2 The concept of voice capacity in the current context is not applicable to 1xEV-DO systems.
3 Traffic configurations requiring voice users are not applicable to 1xEV-DO systems.
3GPP2 C.R1002-A v1.0
97
process is defined per sector. Arrival of a mobile is accounted as a part of the arrival 1
process of the serving sector of that mobile. The sector with the smallest total path 2
loss is the serving sector of the mobile. 3
4.2.2 TCP Modeling 4
The TCP connection set-up and release protocols use a three-way handshake mechanism 5
as described in Figure 4.2.2-1. 6
After the call setup process is completed at time t=t0+580ms, the procedure for MS to set 7
up a TCP session is as follows: 8
1. MS sends a 47-byte1 SYNC packet and wait for ACK from remote server. 9
2. MS starts TCP in slow-start mode (The ACK flag is set in the first TCP segment). 10
The MS starts this process by generating the first TCP packet for FTP transfer. The 11
ACK flag is embedded in this TCP packet along with FTP data. 12
13
The procedure for MS to release the TCP session is as follows: 14
1. MS sets the FIN flag in the last TCP segment. The FIN flag is embedded in the 15
TCP packet along with FTP data. 16
2. MS receives ACKs for all TCP segments from the remote server and terminates the 17
session (MS leaves the system). 18
19
1 The TCP/IP header of 40 bytes + 7 bytes PPP framing overhead = 47 bytes for the SYNC packet.
3GPP2 C.R1002-A v1.0
98
MS BS Router
SYNC
ACK (sets ACK flag in the first TCP segment)
FTP File Transfer
FIN (sets FIN flag in the last TCP segment)
T=t0
MS starts
call setup
T=t0+580ms
MS arrival
time
MS leaves
the system
SYNC/ACK
ACK
Server
1
Figure 4.2.2-1: Modeling of TCP three-way handshake 2
The amount of outstanding data that can be sent without receiving an acknowledgement 3
(ACK) is determined by the minimum of the congestion window size of the transmitter and 4
the receiver window size. After the connection establishment is completed, the transfer of 5
data starts in slow-start mode with an initial congestion window size of 1 segment. The 6
congestion window increases by one segment for each ACK packet received by the sender 7
regardless of whether the packet is correctly received or not, and regardless of whether the 8
packet is out of order or not. This results in an exponential growth of the congestion 9
window. This process is illustrated in Figure 4.2.2-2. 10
3GPP2 C.R1002-A v1.0
99
rt
Client BS Router Server
Access Link Connecting Link
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
r
rt
rt
1
Figure 4.2.2-2: TCP Flow Control During Slow-Start; l = Transmission Time over the 2
Access Link (RL); rt = Roundtrip Time 3
3GPP2 C.R1002-A v1.0
100
The round-trip time in Figure 4.2.2-2, rt, consists of two components: 1
rt = cr + lr (4.2.2-1) 2
where cr = the sum of the time taken by a TCP data segment to travel from the base station 3
router to the server plus the time taken by an ACK packet to travel from the server to the 4
client; lr = the transmission time of a TCP data segment over the access link from the client 5
to the base station router. cr is further divided into two components; 2 = the time taken by 6
a TCP data segment to travel from the base station router to the server plus the time taken 7
by an ACK packet to travel from the server back to the base station router and 3 = the time 8
taken by the ACK packet to travel from the base station router to the client. 9
cr (=2 +3), which accounts for the delay on the connecting link and the TCP ACK 10
transmission on the F-PDCH is modeled as sum of two random variables, 2 is modeled as 11
an exponentially distributed random variable while the time to transmit a TCP ACK on the 12
F-PDCH (3) is modeled as a lognormal distributed random variable with the same mean 13
and standard deviation, as defined in 4.2.4.3. The values for the different delay 14
components are given in Table 4.2.2-1. 15
Table 4.2.2-1 Delay components in the TCP model for the RL upload traffic 16
Delay component Symbol Value
The transmission time of a TCP data
segment over the access link from the
client to the base station router.
1r Determined by
the access link throughput
The sum of the time taken by a TCP
data segment to travel from the base
station router to the server and the
time taken by an ACK packet to travel
from the server to the base station
router.
2 Exponential distribution
Mean=50ms.
The time taken by a TCP data
segment to travel from the base
station router to the client.
3 Lognormal distribution
Mean = 50ms
Standard deviation=50ms
17
From Figure 4.2.2-2, it can be observed that, during the slow-start process, for every ACK 18
packet received by the sender two data segments are generated and sent back to back. 19
Thus, at the mobile station, after a packet is successfully transmitted, two segments arrive 20
back-to-back after an interval cr = 2 + 3. Based on this observation, the packet arrival 21
process at the mobile station for the upload of a file is shown in Figure 4.2.2-3. It is 22
described as follows: 23
3GPP2 C.R1002-A v1.0
101
1. Let S = size of the FTP upload file in bytes. Compute the number of packets in the file, N 1
= S/(MTU-40). Let W = size of the initial congestion window of TCP1. The MTU size is 2
fixed at 1500 bytes 3
2. If N>W, then W packets are put into the queue for transmission; otherwise, all packets 4
of the file are put into the queue for transmission in FIFO order. Let P=the number of 5
packets remaining to be transmitted beside the W packets in the window. If P=0, go to 6
step 6 7
3. Wait until a packet of the file in the queue is transmitted over the access link 8
4. Schedule arrival of next two packets (or the last packet if P=1) of the file after the packet 9
is successfully ACKed. If P=1, then P=0, else P=P-2 10
5. If P>0 go to step 3 11
6. End. 12
1 Compressed TCP header (5 bytes from 40 bytes) and PPP framing overhead (7
bytes + ceiling (size(payload + compressed TCP header + 6)/128)) shall be
transmitted over the air in addition to the payload (MTU – 40) bytes.
For the 1500 MTU size:
1460 Payload (MTU – 40)
+ 5 Compressed TCP header
+ 7 PPP headers and CRC
+ 12 PPP escaping = ceiling ( (1460 + 5 + 6)/128)
1484 bytes above the multiplex sublayer
3GPP2 C.R1002-A v1.0
102
S = size of f ile in by tes
N = CEIL(S/(MTU-40));
W = initial TCP windowsize;
C = min(N,W);
Enqueue C packets f or
transmission;P=N-C;
y es
no
t = current time;
Schedule arriv al of
min(P,2) next packets
of the f ile at the queue
at T=t+tcr
;
P = P - min(P,2)
P > 0 ?
Wait until a packet of
the f ile in the queue is
transmitted
End
1
Figure 4.2.2-3 Packet Arrival Process at the mobile Station for the Upload of a File 2
Using TCP 3
4.2.3 FTP Upload / Email 4
Since FTP uses TCP as its transport protocol, the TCP traffic model described in Section 5
4.1.2 is used to represent the distribution of TCP packets for the FTP upload traffic on the 6
RL. 7
The file upload and email attachment upload are modeled as in Table 4.2.3-1. 8
9
3GPP2 C.R1002-A v1.0
103
Table 4.2.3-1: FTP Characteristics 1
Arrival of new users Poisson with parameter
Upload file size
Truncated lognormal; lognormal pdf:
21 lnexp , 0
22 2
2.0899, 0.9385
xf xx
x
Min = 0.5 kbytes
Max = 500 kbytes
If the value generated according to the
lognormal pdf is larger than Max or
smaller than Min, then discard it and
regenerate a new value.
The resulting truncated lognormal
distribution has a mean = 19.5 kbytes and
standard deviation = 46.7 kbytes
The FTP traffic is simulated as follows: 2
At the beginning of the simulation there are 5 FTP users1 waiting to transmit. 3
o Before transmitting, call setup is performed for each user 4
5
Afterwards, FTP upload users arrive according to the Poisson arrival process, as defined in 6
Table 4.2.3-1. 7
For each new FTP upload user coming into the system, call setup is performed 8
Each FTP upload user stays in the system until it finishes the transmission of its 9
file 10
After an FTP upload user finishes the transmission of its file, it immediately leaves 11
the system. 12
Since the arriving FTP users are dropped uniformly over 19 cells, it is possible the number 13
of users can exceed the sector capacity. In that case, the new arrival should be blocked. 14
The blocking rate should be recorded. 15
16
4.2.4 HTTP Model 17
The following figure is an example of events occurring during a HTTP session. 18
1 In order to skip the transient period, the number of 5 initial FTP users is taken to represent the
number of users at steady state.
3GPP2 C.R1002-A v1.0
104
Page
request
generated
Page
request
sent
TCP ACK
generated
TCP ACK
sent
TCP segment of the
requested page
received at BS
BS
Transmission
Timeline
MS Timeline
Reading
time
Internet
delay
TCP segment
sent on FL
Internet
delay
Requests for
embedded
objects sent
TCP segments
of the requested
objects
received at BS
Internet
delay
RL
delay
Parsing
time
Requests for
embedded
objects
generated
New TCP
segments
received at BS
Main Object
FL delay
FL
delay
RL
del
ay
RL
delay
FL
delay
1
Figure 4.2.4-1: Example of events occurring during web browsing. 2
4.2.4.1 HTTP Traffic Model Parameters 3
Reading time (Dpc): modeled as in Table 4.2.4-1 4
Internet delay (DI): modeled as an exponentially distributed random variable with a 5
mean of 50ms 6
Parsing time (Tp): modeled as in Table 4.2.4-1 7
RL delay: specific for the implemented system. Includes RL packet transmission 8
delay and scheduling delay (if scheduled) 9
FL delay (DFL): defined as the time a TCP segment is first in the queue for 10
transmission until it finishes transmission on forward link. The delay includes 11
transmission delay and forward link scheduling delay. If there are multiple packets, 12
each packet has its own additional contribution to the overall DFL. The model is 13
given in 4.2.4.3. 14
Number of TCP segments in the main object (NM). NM = SM /(MTU-40). The main 15
object size, SM, is generated according to Table 4.2.4-1 16
Number of TCP segments in embedded object (NE). NE = SE /(MTU-40). The 17
embedded object size, SE, is generated according to Table 4.2.4-1 18
Number of embedded objects (Nd). Modeled according to Table 4.2.4-1 19
HTTP1.1 mode 20
3GPP2 C.R1002-A v1.0
105
The opening and the closing of the TCP connections is not modeled1 1
HTTP request size = 350 bytes 2
Requests for embedded objects are pipelined – all requests are buffered together 3
MTU size = 1500 bytes 4
ACK size = 12 bytes2 5
Every received TCP segment is acknowledged. 6
1 This does not have much influence since in HTTP1.1 persistent TCP connections are used to
download the objects (located at the same server) and the objects are transferred serially over a single
TCP connection.
2 Compressed TCP/IP header (5 bytes from 40 bytes) and HDLC framing and PPP overhead (7 bytes).
3GPP2 C.R1002-A v1.0
106
Table 4.2.4-1: HTTP Traffic Model Parameters 1
Component Distribution Parameters PDF
Main object
size (SM)
Truncated
Lognormal
Mean = 9055 bytes
Std. dev. = 13265
bytes
Minimum = 100 bytes
Maximum = 100
Kbytes
If x > max or x < min,
then discard and re-generate a new value for
x.
21 lnexp , 0
22 2
1.37, 8.35
xf xx
x
Embedded
object size
(SE)
Truncated
Lognormal
Mean = 5958 bytes
Std. dev. = 11376
bytes
Minimum = 50 bytes
Maximum = 100
Kbytes
21 lnexp , 0
22 2
1.69, 7.53
xf xx
x
If x > max or x < min, then discard and re-
generate a new value for x.
Number of
embedded
objects per
page (Nd)
Truncated
Pareto
Mean = 4.229
Max. = 53
55,2,1.1
,1
mk
mxkx
af k
x
Note: Subtract k from the
generated random value to
obtain Nd
If x > max, then discard and re-
generate a new value for x
Reading
time (Dpc) Exponential Mean = 30 sec
, 0
0.033
xxf ex
Initial
reading time
(Dipc)
Uniform Range [0, 10] s
1,
0, 10
a x bf x b a
a b
Parsing time
(Tp) Exponential Mean = 0.13 sec
, 0
7.69
xxf ex
4.2.4.2 Packet Arrival Model for HTTP 2
At the beginning of the simulation, call setup is performed for all HTTP users. After that, 3
the simulation flow is described as follows: 4
3GPP2 C.R1002-A v1.0
107
1. Generate an initial reading time Dipc.1 Wait Dipc seconds. 1
2. Initiate the TCP window size W=1 2
3. Generate a request for the main page 3
4. Wait for the requests to go through the RL and reach the bases station (RL delay): 4
In case these are requests for embedded objects, wait until all requests reach 5
the base station. 6
5. Generate an Internet delay DI. Wait DI seconds. 7
6. Generate random delays, which define the time instances when each of the TCP 8
segment transmission is completed the FL. The number of these instances is: 9
a. For the main page: 10
i. At the very beginning of the packet call: 1. 11
ii. Afterwards: min(2n, #of outstanding TCP segments on FL), where n is 12
the number of ACKs received in the last physical layer packet (from 13
the step 9.a.i) 14
b. For embedded objects: 15
i. At the very beginning of the transmission of embedded objects: 16
min(W,
dN
i
i
EN1
). 17
ii. Afterwards: min(2n, #of outstanding TCP segments on FL), where n is 18
the number of ACKs received in the last physical layer packet (from 19
the step 9 a.i. 20
7. Every time instance of the completed TCP segment transmission on FL generates an 21
ACK on RL 22
8. Continue RL simulation – when ACK is generated, reduce the number of 23
outstanding TCP packets by 1 24
9. Examine if the transmission of the very last TCP segment of the HTTP object is 25
completed: 26
a. If no: 27
i. Proceed with simulation until next ACK or a group of n ACKs within a 28
single physical layer packet is transmitted 29
ii. Increase W:=W+n 30
iii. Go to step 5 31
b. If yes, for main page: 32
1 The initial reading time is defined differently from subsequent reading times in order to ensure that
all HTTP users finish the reading time within a limited period.
3GPP2 C.R1002-A v1.0
108
i. Generate Tp (parsing time) 1
ii. Generate requests for embedded objects 2
iii. Continue RL simulation - transmit outstanding ACK(s) for the main 3
page and accordingly increment W:=W+n for each group of n ACKs 4
transmitted, until requests for embedded objects are generated 5
iv. Go to step 4 6
c. If yes, for embedded objects: 7
i. Generate Dpc (reading time) 8
ii. Continue RL simulation - transmit outstanding ACK(s) for the 9
embedded objects 10
iii. Go to step 2 when reading time expires or until all ACKs are 11
transmitted, whichever is longer 12
4.2.4.3 Forward Link Delay Model for HTTP Users 13
Forward link delay (DFL) is defined as a time needed for transmission of a TCP segment that 14
is first in the queue for transmission. The delay includes transmission delay and forward 15
link scheduling delay (waiting for other users to use the FTC/F-PDCH). It is modeled 16
according to a distribution obtained from the forward link simulation. 17
The forward link delay is simulated as follows: 18
1. The time to transmit a TCP packet is simulated as a lognormal distributed random 19
variable with the same mean and standard deviation 20
2. The mean and standard deviation of the time to transmit a TCP packet (in PCGs) for 21
a given user has to be computed with the following expression: 22
sec/800*** PCGsPkt
SpktsU
DRC
PST
(1xEV-DV Systems) (4.2.4.3-1) 23
or 24
sec/600*** slotsPkt
SpktsU
DRC
PST
(1xEV-DO Systems) (4.2.4.3-2) 25
where, 26
PS (Packet Size) is 12,000 bits (for MTU=1500 bytes) 27
U (average number of users who have data to transmit) = 6.28 (for data only 28
case) 29
Pkt
Spkts(average number of subpackets per packet) = 1.32 30
3. The ATR (average transmission rate) in the expression above is computed based on 31
the geometry (see below) and channel model of a given user. The DRC for a user is 32
3GPP2 C.R1002-A v1.0
109
obtained by linear interpolation/extrapolation between the points in Table 4.2.4-2 1
for the appropriate channel model. If the DRC resulting from extrapolating between 2
the points in Table 4.2.4-2 is 0 bps, the user is discarded and replaced. 3
Geometry = 1 / ( (Ioc+No)/Ior + ) 4
For geometry computation, all BTSs are assumed to be transmitting at full 5
power (20 W) 6
Ior is the total energy per chip received from the serving sector, which is 7
assumed to be transmitting at max transmit power (20 Watts) 8
Ioc is the sum of total energy per chip from all other sectors, each assumed to be 9
transmitting at max transmit power (20 Watts) 10
No is the thermal noise spectral density 11
Table 4.2.4-2 Points to obtain the average transmission rate (ATR) given the 12
geometry and channel model of a user 13
Channel Model A Channel Model B Channel Model C Channel Model D Channel Model E
Geometry
(dB)
Average
Transmission
Rate (kbps)
Geometry
(dB)
Average
Transmission
Rate (kbps)
Geometry
(dB)
Average
Transmission
Rate (kbps)
Geometry
(dB)
Average
Transmission
Rate (kbps)
Geometry
(dB)
Average
Transmission
Rate (kbps)
-8 200 -7 280 -6 35 -4 150 -5 290
11 2,250 11 1,500 11 1,100 0 330 3 980
NA NA NA NA NA NA 11 1,650 11 2,500
4.2.5 WAP Users 14
Figure 4.2.5-1 illustrates the data flow for the WAP traffic model, and Table 4.2.5-1 15
describes the distribution of the model parameters. 16
3GPP2 C.R1002-A v1.0
110
MS BS
PDSN/WAP
GatewayContent
Server
WAP Gateway
Response Time
(exponential)
Reading Time -
includes browser rendering delay
(exponential)
Request
Response, 1st object
(size is truncated pareto distributed)
HTTP Response
HTTP GET
N = #
objects
per
response
(geometr
ic)
Response, 2nd object
(size is truncated pareto distributed)
object inter-arriv al
time (exponential)
Response, Nth object
(size is truncated pareto distributed)
....
RequestHTTP GET
1
Figure 4.2.5-1: Packet Trace for the WAP Traffic Model 2
During the simulation period, the model assumes that each WAP user is continuously 3
active, i.e., making a WAP request, waiting for the response, waiting the reading time, and 4
then making the next request. 5
WAP traffic on the RL simply represents the WAP requests used by the mobile to download 6
the WAP pages. 7
At the beginning of the simulation, call setup is performed for all WAP users. After that, the 8
simulation flow is as follows: 9
1. Generate the reading time Tri. Wait Tri seconds. 10
2. Generate NW (number of objects per response) 11
3. Initialize N:=NW 12
4. At the end of the reading time generate WAP request of 76 bytes 13
5. Generate the gateway response time Tg. Wait Tg seconds. 14
6. Decrement N:=N-1 15
7. Examine if N > 0 16
o If yes 17
3GPP2 C.R1002-A v1.0
111
Generate an object inter-arrival time TIA. Wait TIA seconds. 1
Decrement N:=N-1 2
Go to step 7 3
o If no 4
Wait Tr (reading time) 5
Go to step 2 6
Table 4.2.5-1: WAP Traffic Model Parameters 7
Packet based
information types
Size of WAP request
Object size
# of objects
per response
NW
Inter-arrival time
between objects Tia
WAP gateway response time Tg
Reading time Tr
Initial reading time Tri
Distribution
Deterministic Truncated Pareto
(Mean= 256
bytes, Max= 1400 bytes)
Geometric P(k) =
0.5k
,
k>=1
Exponential Exponential Exponential Uniform
Distribution Parameters
76 octets K = 71.7 bytes, = 1.1
Mean = 2 Mean = 1.6 s
Mean = 2.5 s
Mean = 5.5 s
Range [0, 5] s
4.2.6 Reverse Link Delay Criteria for HTTP/WAP 8
Reverse link delay for a TCP ACK, defined as the time from the moment the TCP ACK is 9
generated at the mobile station until it is received at the base station, shall not be more 10
than 160 ms 90% of the time. 11
Reverse link delay for a HTTP request, defined as the time from the moment the HTTP 12
request is generated at the mobile station until it is received at the base station, shall 13
satisfy the following criteria. The CDF curve of the reverse link HTTP request delays shall lie 14
to the left of the curve given by the three points in Table 4.2.6-1. 15
Table 4.2.6-1 Reverse link delay criteria for HTTP request 16
HTTP request delay [s] CDF
1.0 0.8
0.6 0.5
0.4 0.2
Reverse link delay for a WAP request, defined as the time from the moment the WAP 17
request is generated at the mobile station until it is received at the base station, shall not 18
be more than 300 ms 90% of the time. 19
3GPP2 C.R1002-A v1.0
112
4.2.7 Mobile Network Gaming Model 1
This section describes a model for mobile network gaming traffic on the reverse link. Table 2
4.2.7-1 describes the parameters for the mobile network gaming traffic on the reverse link. 3
Table 4.2.7-1 Mobile network gaming traffic model parameters 4
Component Distribution PDF and generation method
Initial
packet
arrival
Uniform (a=0,
b=40ms) bxa
abxf
1)(
Packet
arrival
Deterministic
(40ms)
Packet size Extreme
(a=45 bytes, b
= 5.7)
0,1
)(
beeb
xfb
ax
eb
ax
ln lnX a b Y , )1,0(UY
Because packet size has to be integer
number of bytes, the largest integer less
than or equal to X is used as the actual
packet size.
UDP header Deterministic
(2 bytes)
5
This model uses Largest Extreme Value distribution for the packet size. For cellular system 6
simulation, 2-byte UDP header (after header compression) should be added to the packet 7
size X . Because packet size has to be integer number of bytes, the largest integer less than 8
or equal to X is used as the actual packet size. To simulate the random timing 9
relationship between client traffic packet arrival and reverse link frame boundary, the 10
starting time of a network gaming mobile is uniformly distributed within [0, 40ms]. 11
A maximum delay of 160ms is applied to all reverse link packets, i.e., a packet is dropped 12
by the mobile station if any part of the packet have not started physical layer transmission, 13
including HARQ operation, 160ms after entering the mobile station buffer. A packet can 14
start physical layer transmission at the 160ms time instant. Packet dropping should be the 15
last operation of mobile station buffer management, if any, at any time instant. The packet 16
delay of a dropped packet is counted as 180ms. 17
A mobile network gaming user is in outage if the average packet delay is greater than 60ms. 18
The average delay is the average of the delay of all packets, including the delay of packets 19
delivered and the delay of packets dropped. 20
3GPP2 C.R1002-A v1.0
113
4.2.8 Voice Model (1xEV-DV Systems Only) 1
Voice follows the standard Markov source model. The voice activity factor is 0.403 with 29% 2
full rate, 60% eighth rate, 4% half rate, and 7% quarter rate. The corresponding transition 3
probabilities are defined in C.S0025 (TIA/EIA/IS-871). For the received frame in the system 4
level simulation of voice traffic, lookup_Eb/Nt(rate) is computed in order to lookup the FER 5
on the 9600 bps FER versus Eb/Nt curve for all 4 rates. 6
Lookup_Eb/Nt(rate) is computed as follows: 7
lookup_Eb/Nt(rate) = T/P(9600) * 1228800/9600 * Ec/Nt(pilot) 8
where Ec/Nt(pilot) is the average pilot power received over the frame. 9
10
4.3 Common Traffic Models Applicable for Both Forward Link and Reverse Link 11
Services 12
4.3.1 Voice over IP(VoIP) 13
4.3.1.1 Source Configuration Files 14
For each simulation run a source configuration file is defined which lists the set of AT‘s 15
used in the simulation. For each AT the file specifies an AT layout number, a traffic type, 16
and slot offsets for the audio files. The AT layout number references an entry in the AT 17
layout file. Traffic type specifies one of full buffer, VoIP, or VT. The audio offset field is set to 18
-1 for full buffer traffic. Otherwise it specifies the point to begin reading in the audio source 19
file for the specified AT at the start of the simulation. 20
The sample source configuration file is specified by [36]. In this file, the audio offsets were 21
generated using a uniform random number in the range [1, NumSlots] in each source file. 22
Appendix U-1 details the format of source configuration files. 23
4.3.1.2 Source Files 24
One source file is specified for VoIP which is used by each AT with a unique starting offset 25
(specified in the source configuration file) in the file for each AT. 26
The source files are to be interpreted as circular, in the sense that as the end of the file is 27
reached, the source continues from the beginning of the file. The exact wraparound point 28
is specified as a parameter in the file for the total number of slots the file represents 29
(referred to as NumSlots). After this final slot number is reached while scanning through 30
the file, the file wraps around to the beginning, and the next slot scanned should be slot 1. 31
The first line of each source file specifies the number of lines describing arrival packets in 32
the file (referred to as NumLines, provided for convenience in scanning the file). The second 33
line (NumSlots) defines the total number of slots the file represents. After NumSlots, there 34
are NumLines remaining lines in the file, and each represents one application packet 35
arrival. Each such line contains three fields, the arrival slot, the packet size, and the 36
packet type. The arrival slot indicates the offset from the beginning of the file (in slots) of 37
the arrival time for the packet. The packet size indicates the size of the application packet, 38
3GPP2 C.R1002-A v1.0
114
in bytes. The meaning of the packet type field is source-specific. For the audio source, it is 1
always set to 0. 2
Appendix U-2 details the format of source files (see [37]). The audio file was generated 3
based on the MSO model IS-871, but with some alterations. Modeling of 1/8th frame rate 4
blanking is achieved by only transmitting the first 1/8th rate frame of each silence interval 5
(as a silence indicator), and then one out of every 12 consecutive 1/8th rate frames (see 6
[39] for a description of this approach to model blanking). We assume a 4-byte RoHC 7
overhead for each IP packet, and this is included in the size of each VoIP packet in the 8
source file. 9
VoIP delay jitter model is applied for the generation of source files for the forward link VoIP 10
simulation. Laplacian distribution with 0 and 11.5 ms is used to model VoIP delay 11
jitter. 12
XXF exp
2
1)( , for X , and 13
XXF exp
2
11)( , for X . (4.3.1.2-1) 14
For a voice frame generated at T, the corresponding VoIP packet arrives at AN equals T + τ, 15
where τ ~ L(0, 5.11ms) with limit msms 8080 . The VoIP source file with delay jitter 16
applied is available as [42]. 17
4.3.1.3 Simulation Specifics 18
The VoIP simulation is to be run for 60k slots, with 10k warm-up slots. Units of the 19
simulation are specified in DO slots (i.e. 5/3 msec). Information used for packet scheduling 20
and dropping is to be what is actually available in the specified design. Any packet may be 21
dropped on both FL and RL at any point in the simulation, but statistics on dropped and 22
lost packets are collected for each AT. 23
When presenting simulation results, parameters configurable in the system should be 24
summarized, such as MAC and QoS parameters. 25
Each link (FL and RL) is simulated separately. The simulation flow is as follows: 26
1. Drop a number of users (K) per sector. 27
2. For each user perform server selection and redrop the user to another location if 28
either FL/RL server selection is unsuccessful (i.e. the user is on either FL/RL coverage 29
outage). 30
3. For each user (e.g. the k-th user), find the delay that corresponds to the 98-th 31
percentile of the user's packet delay cdf - this is denoted by D1. 32
4. Store all users' D1 values and plot the cdf. 33
5. Find the largest K that has the 95-th percentile of the cdf less than the delay 34
criterion D0 by increasing K (go to step 1). 35
3GPP2 C.R1002-A v1.0
115
6. Report the capacity as the number of users K0 that satisfied the step 5. 1
The values of D0 to be used are: 2
D0_FL: 50 ms, 70 ms 3
D0_RL: 50 ms, 70 ms 4
To avoid hunting for the exact integer value of K, it is assumed that K takes on values 10n 5
where n is integer. 6
7
4.3.1.4 VoIP Statistics 8
Application Packet Erasure, Drop, and Success 9
An application packet sees one of three possible outcomes: erasure, drop, or success. An 10
application packet erasure occurs when any portion of the packet is lost on the channel, 11
after any H-ARQ or other re-transmissions have failed. An application packet drop occurs 12
when no portion of the packet is ever placed in a physical layer packet for transmission. An 13
application packet success occurs when the entire packet is successfully transmitted. 14
At the end of the simulation run, application packets that have any portion of their data 15
still in queue or in flight (i.e. not yet successfully delivered) are considered dropped. 16
Application Packet Delay 17
The delay for a successfully delivered application packet is the time between packet arrival 18
and successful transmission of the last bit in the packet. For erased or dropped application 19
packets, delay is defined to be infinite. 20
Loss and Drop cdf’s 21
Statistics on proportion of dropped and erased packets are to be kept separately for audio 22
packets for each AT (that is, in proportion to total number of packets of that type). A cdf of 23
each of the audio dropped packet rates over all AT‘s is to be provided, and similarly for 24
erased packets for each of these three cases. The required full-network results of a VoIP 25
simulation are summarized in Appendix E. 26
Audio Tail Delay 27
For each AT, determine the audio tail delay point as follows. For audio, find the smallest 28
delay value Da such that TargetAudFER percent or less of the audio packets have delay 29
equal to or longer than Da. If no finite Da exists that satisfies this criterion (i.e. more than 30
TargetAudFER percent of the audio packets are dropped or erased), then Da is defined to be 31
infinite. This delay value Da is the audio tail delay associated with that specific AT, and 32
each AT has exactly one such value. We use the following value for VoIP for both FL and RL 33
scenarios: TargetAudFER = 2%. 34
A cdf of the value of the audio tail delay over all VoIP AT‘s is to be provided as part of the 35
VoIP simulation results. The required full-network results of a VoIP simulation are 36
summarized in Appendix U-4. 37
Per-AT Result Data 38
3GPP2 C.R1002-A v1.0
116
An ASCII text file is to be generated for each VoIP simulation run, which contains one line 1
for each AT that runs VoIP in the simulation. The VoIP statistics described above are to be 2
included for each such AT. Details on the format of the per-AT result data can be found in 3
Appendix U-3. Note that all of the network results required in Appendix E can be derived 4
from this per-AT data file. 5
4.3.1.5 Source Mix 6
For simulations containing only VoIP sources, each AT either transmits or receives an 7
audio flow. When running the mixed VoIP and full-buffer simulation, the statistics as 8
described here for VoIP are to be taken for the VoIP flows. Full-buffer throughput and the 9
fairness criterion described in 2.1.6.2 are to be used for the results of the full-buffer flows. 10
4.3.1.6 Scheduler Statistics 11
Statistics appropriate to the scheduler used shall be reported, e.g. scheduling delay, 12
fraction of utilized BW, and inter-packet departure time. 13
14
4.3.2 Video Telephony(VT) 15
4.3.2.1 Source Configuration Files 16
For each simulation run a source configuration file is defined which lists the set of AT‘s 17
used in the simulation. For each AT the file specifies an AT layout number, a traffic type, 18
and slot offsets for the audio and video source files. The AT layout number references an 19
entry in the AT layout file. Traffic type specifies one of full buffer, VoIP, or VT. The audio 20
offset field is set to -1 for full buffer traffic. Otherwise it specifies the point to begin reading 21
in the audio source file for the specified AT at the start of the simulation. Similarly, the 22
video offset field is set to -1 for full buffer and VoIP traffic. Otherwise it specifies the point to 23
begin reading in the video source file for the specified AT at the start of the simulation. 24
A sample source configuration file is specified by [36]. The file is for a mixed VT/full buffer 25
simulation. In this file, the audio and video source offsets were generated using a uniform 26
random number in the range [1, NumSlots] in each source file. Appendix U-5 details the 27
format of source configuration files. 28
4.3.2.2 Source Files 29
Two source files are specified for VT, one for audio and one for video. The file for audio is 30
the same file as that used for VoIP. For VT each AT has two source flows, one video and one 31
audio, represented by these two files. Each AT uses the same pair of source files, but each 32
AT uses a unique starting point offset in each file, as specified in the source configuration 33
file. 34
The source files are to be interpreted as circular, in the sense that as the end of the file is 35
reached, the source continues from the beginning of the file. The exact wraparound point 36
is specified as a parameter in the file for the total number of slots the file represents 37
3GPP2 C.R1002-A v1.0
117
(referred to as NumSlots). After this final slot number is reached while scanning through 1
the file, the file wraps around to the beginning, and the next slot scanned should be slot 1. 2
The first line of each source file specifies the number of lines describing arrival packets in 3
the file (referred to as NumLines, provided for convenience in scanning the file). The second 4
line (NumSlots) defines the total number of slots the file represents. After NumSlots, there 5
are NumLines remaining lines in the file, and each represents one application packet 6
arrival. Each such line contains three fields, the arrival slot, the packet size, and the 7
packet type. The arrival slot indicates the offset from the beginning of the file (in slots) of 8
the arrival time for the packet. The packet size indicates the size of the application packet, 9
in bytes. The meaning of the packet type field is source-specific. For the audio source, it is 10
always set to 0. For the video source, packet type specifies whether the packet is an I-11
frame or a P-frame, where 0 means P-frame and 1 means I-frame. 12
Appendix U-2 details the format of source files. A pair of source files is provided for the VT 13
simulations (see [37] and [38]). The audio file was generated based on the MSO model IS-14
871, but with some alterations. Modeling of 1/8th frame rate blanking is achieved by only 15
transmitting the first 1/8th rate frame of each silence interval (as a silence indicator), and 16
then one out of every 12 consecutive 1/8th rate frames (see [39] for a description of this 17
approach to model blanking). 18
The video file was generated using an H.263 encoder on reference video clips (see Appendix 19
F for a detailed description of the source video file). The source file [37] was created in the 20
following manner. A set of reference video clips (detailed in [40]) was encoded under a few 21
different reasonable assumptions to create a set of reference encodings. The video clips 22
used were: crossing, doctor, foreman, friends, stunt, walk, zoom (7 of them). Two rate 23
types of encoding were used: fixed rate at 44kbps, and fixed quality with QP fixed to a 24
value of 22. Four encoding modes were used, varying fps and the GOP (i.e. the rate of I-25
frames vs. P-frames): 1) 15fps, GOP 45, 2) 15fps, GOP 37, 3) 10fps, GOP 27, 4) 10fps, GOP 26
23. Altogether, then, there are 7 * 2 * 4 = 56 separate reference encodings. The source file 27
was generated from these 56 encodings by concatenating a completely random set of them 28
(i.e. each subsequent clip is chosen uniformly from the set of 56 clips). 29
We assume a 4-byte RoHC overhead for each IP packet, and this is included in the size of 30
each application packet in the source file. We account for IP fragmentation in the source 31
by assuming a maximum IP packet payload size of 1460 bytes, and adding in the 4-bytes of 32
RoHC overhead per IP packet needed to carry the application packet. The sum of these 33
overheads are included in the application packet size in the source file. 34
The same VoIP delay jitter model is applied for the generation of audio source file in the 35
forward link simulation. 36
37
4.3.2.3 Simulation Specifics 38
The VT simulation is to be run for 60k slots, with 10k warmup slots. Units of the 39
simulation are specified in DO slots (i.e. 5/3 msec). Information used for packet scheduling 40
and dropping is to be what is actually available in the specified design. Any packet may be 41
3GPP2 C.R1002-A v1.0
118
dropped on both FL and RL at any point in the simulation, but statistics on dropped and 1
lost packets are collected for each AT. 2
When presenting simulation results, parameters configurable in the system should be 3
summarized, such as MAC and QoS parameters. 4
4.3.2.4 VT Statistics 5
Application Packet Erasure, Drop, and Success 6
An application packet sees one of three possible outcomes: erasure, drop, or success. An 7
application packet erasure occurs when any portion of the packet is lost on the channel, 8
after any H-ARQ or other re-transmissions have failed. An application packet drop occurs 9
when no portion of the packet is ever placed in a physical layer packet for transmission. An 10
application packet success occurs when the entire packet is successfully transmitted. At 11
the end of the simulation run, application packets that have any portion of their data still 12
in queue or in flight (i.e. not yet successfully delivered) are considered dropped. 13
Application Packet Delay 14
The delay for a successfully delivered application packet is the time between packet arrival 15
and successful transmission of the last bit in the packet. For erased or dropped application 16
packets, delay is defined to be infinite. 17
Loss and Drop cdf’s 18
Statistics on proportion of dropped and erased packets are to be kept separately for audio, 19
I-frame, and P-frame packets for each AT (that is, in proportion to total number of packets 20
of that type). A cdf of each of the audio, I-frame, and P-frame dropped packet rates over all 21
AT‘s is to be provided, and similarly for erased packets for each of these three cases. The 22
required full-network results of a VT simulation are summarized in Appendix U-7. 23
Audio and Video Tail Delay 24
For each AT, determine the audio and video tail delay point as follows. Two delay 25
parameters are defined, TargetAudFER and TargetVidFER (in percent). For audio, find the 26
smallest delay value Da such that TargetAudFER percent or less of the audio packets have 27
delay equal to or longer than Da. If no finite Da exists that satisfies this criterion (i.e. more 28
than TargetAudFER percent of the audio packets are dropped or erased), then Da is defined 29
to be infinite. This delay value Da is the audio tail delay associated with that specific AT, 30
and each AT has exactly one such value. 31
Similarly, for video find the smallest delay value Dv such that TargetVidFER percent or less 32
of the video packets have delay equal to or longer than Dv. If no finite Dv exists that 33
satisfies this criterion (i.e. more than TargetVidFER percent of the video packets are 34
dropped or erased), then Dv is defined to be infinite. This delay value Dv is the video tail 35
delay associated with that specific AT, and each AT has exactly one such value. Note that 36
Dv is determined from the combined flow of I and P-frames and is not determined 37
separately for each. 38
We use the following values for VT for both FL and RL scenarios: TargetAudFER = 2%, and 39
TargetVidFER = 2%. 40
3GPP2 C.R1002-A v1.0
119
A cdf of the value of the audio tail delay over all VT AT‘s, and a cdf of the value of the video 1
tail delay over all VT AT‘s are to be provided as part of the VT simulation results. Also, a 2
cdf of the value of the difference in audio and video tail delay over all VT AT‘s is to be 3
provided. The required full-network results of a VT simulation are summarized in Appendix 4
U-7. 5
Per-AT Result Data 6
7
An ASCII text file is to be generated for each VT simulation run, which contains one line for 8
each AT that runs VT in the simulation. The VT statistics described above are to be 9
included for each such AT. Details on the format of the per-AT result data can be found in 10
Appendix U-5. Note that all of the network results required in Appendix U-6 can be derived 11
from this per-AT data file. 12
4.3.2.5 Source Mix 13
For simulations containing only VT sources, each AT either transmits or receives both an 14
audio and a video flow. When running the mixed VT and full-buffer simulation, the 15
statistics as described here for VT are to be used for the VT flows. Full-buffer throughput 16
and the fairness criterion described in 2.1.6.2 are to be used for the results of the full-17
buffer flows. 18
4.3.2.6 Scheduler Statistics 19
Statistics appropriate to the scheduler used shall be reported, e.g. scheduling delay, 20
fraction of utilized BW, and inter-packet departure time. 21
22
3GPP2 C.R1002-A v1.0
120
This page intentionally left blank. 1
2
3GPP2 C.R1002-A v1.0
121
APPENDIX A: LOGNORMAL DESCRIPTION 1
The attenuation between a mobile and the transmit antenna of the i-th cell site, or between 2
a mobile and each of the receiving antennas of the i-th cell site is modeled by 3
21010 i
X
ioi RDkLi
(A-1)
where iD is the distance between the mobile and the cell site, is the path loss exponent 4
and iX represents the shadow fading which is modeled as a Gaussian distributed random 5
variable with zero mean and standard deviation . iX may be expressed as the weighted 6
sum of a component Z common to all cell sites and a component iZ that is independent 7
from one cell site to the next. Both components are assumed to be Gaussian distributed 8
random variables with zero mean and standard deviation independent from each other, 9
so that 10
ii bZaZX such that 122 ba (A-2)
Typical parameters are 9.8 and 2
122 ba for 50% correlation. The correlation is 11
0.5 between sectors from different cells, and 1.0 between sectors of the same cell. 12
13
3GPP2 C.R1002-A v1.0
122
APPENDIX B: ANTENNA ORIENTATION 1
Antenna Bearing is the angle between the main antenna lobe center and a line directed due 2
east given in degrees. The Bearing Angle increases in a clockwise direction. Figure B-1 3
below shows the 3-sector 120-degree center cell site with a sector 1 bearing angle of zero 4
degrees. 5
East
0, 120, 240 Degree Sector Orient at ion
120
degreesMain Antenna Lobe
Sector 1
0
120
240
6
Figure B-1 Center Cell Antenna Bearing Orientation diagram 7
8
9
10
11
12
Figure B-2 Orientation of the Center Cell Hexagon 13
Figure B-2 shows the orientation of the center cell hexagon corresponding the antenna 14
bearing orientation diagram of Figure B-1. The main antenna lobe center directions shall 15
point to the sides of the hexagon. The main antenna lobe center directions of the other 18-16
surrounding cell shall be parallel to those of the center cell. 17
Antenna downtilt is the angle between the main antenna lobe center and a line directed 18
perpendicular from the antenna face. As the downtilt angle increases (positively) the 19
antenna main lobe points increasingly toward the ground. 20
The Antenna gain that results for a given bearing and downtilt angle for a given cell/sector 21
'i' with respect to a mobile 'k' is characterized by the equations given below (B-1,2). A 22
geometric representation is given in Figure B-3 below. 23
(i,k) = (i,k) - (i) (B-1) 24
(i,k) = (i,k) - (i,k) (B-2) 25
where 26
(i,k) = angle between antenna main lobe center and line connecting cell 'i' 27
and mobile 'k' in radians in horizontal plane. 28
3GPP2 C.R1002-A v1.0
123
(i,k) = angle between antenna main lobe center and line connecting cell 'i' 1
and mobile 'k' in radians in vertical plane. 2
(i) = antenna bearing for cell 'i' in radians. 3
h(i) = antenna height for cell 'i' in meters. 4
(i) = antenna downtilt in radians . 5
(i,k) = corrected downtilt angle for given horizontal offset angle ((i,k)) in radians. 6
= ATAN( COS((i,k))TAN((i)) ) 7
(i,k) = antenna-mobile line of site angle in radians 8
= ATAN( h(i)/d(i,k) ) . 9
d(i,k) = (mobile(k)_ypos - cell[i]_ypos)2 + (mobile(k)_xpos - cell[i]_xpos)2 = distance 10
(i,k) = mobile bearing is the angle between the line drawn between the cell and mobile 11
and a line directed due east from the cell. 12
= ATAN2( (mobile(k)_ypos - cell[i]_ypos, (mobile(k)_xpos - cell[i]_xpos) ) 13
sector 'i'
East
0
x
mobile 'k'
Antenna Main Lobe
Horizontal Angle ( ) off Main Lobe center
h
d mobile 'k'
o
Vertical Angle ( ) off Main Lobe center
Main Lobe Center
sector 'i'
14
15
Figure B-3 Mobile Bearing orientation diagram example. 16
17
3GPP2 C.R1002-A v1.0
124
APPENDIX C: DEFINITION OF SYSTEM OUTAGE AND VOICE CAPACITY 1
The voice capacity Nmax is defined as the maximum number of voice users that a sector can 2
support while the system outage is below certain probability. System outage for voice users 3
is to be evaluated based on the percentage of voice users in per-link outage. To get the per-4
link forward or reverse outage for a given voice user, one shall evaluate the short-term FER 5
for a voice link by measuring the FER over windows of 400 ms (20 20-ms frames). 6
The proposed DV systems shall not be in system outage for more than 3% of the time. This 7
is defined as follows: 8
Assume the short-term FER for user i is FERi,j, where i = 1,…,N, is the user index for all 9
cells and j=1,…, M, is the time index spanning a simulation run. 10
Tsystem outage, the system outage target, is the limit: 11
Prob.(Per-user outage among all N users in all runs) < Tsystem outage = 3% 12
Per-user outage is defined as the event where a user‘s voice connection in either direction 13
has short-term FER higher than 15% more often than Tper link = 1% of the time. That is, 14
(Per-user outage for user i) = [ (j=1 to M (Ii,j)/M > 1% for either forward or reverse] 15
where the indicator function Ii,j is defined by: Ii,j = 1 if FERi,j >= 15% and 0 if FERi,j < 15% 16
Note that the actual user in outage would perceive the quality to be low when either 17
forward or reverse link is in outage. 18
To simplify the simulations, no voice calls will be dropped during each simulation run and 19
the same number of voice calls is maintained throughout each simulation run. 20
21
3GPP2 C.R1002-A v1.0
125
APPENDIX D: FORMULA TO DEFINE VARIOUS THROUGHPUT AND DELAY 1
DEFINITIONS 2
For each fixed simulation condition (e.g., voice load, distribution of different traffic, number 3
of data users, etc.), simulation is run for multiple independent runs using Monte Carlo 4
approach. Let 5
T = simulation time. 6
M = total number of independent runs (for a specific configuration). 7
N= total number of data users in each run (for a sector). 8
m= index of the simulation runs, i.e., m = 1,2,3, …, M. 9
n = index of a data user within a simulation run, i.e., n = 1,2,…, N. 10
Therefore, the n-th data user in the m-th simulation run can be specified by user(m,n). 11
Let 12
K(m,n) = total number of packet calls generated for user(m,n). 13
k = index of packet calls for a user. For user(m,n),, k = 1, 2, …, K(m,n). 14
L(m,n,k) = total number of packets generated for the k-th packet call of user(m,n). 15
l = index of packet within a packet call. For the k-th packet call of user(m,n), l = 1,2, 16
…, L(m,n,k). 17
B(m,n,k,l) = number of information bits contained in the l-th packet of the k-th 18
packet calls for user(m,n) . If the packet is not successfully delivered by the end of 19
the simulation run, B(m,n,k,l) = 0. 20
TA(m,n,k,l) = arrival time of the l-th packet of the k-th packet calls for user(m,n). it 21
is the time when the packet arrives at the transmitter side and is put into a queue. 22
TD(m,n,k,l) = delivered time of the l-th packet of the k-th packet calls for user(m,n) . 23
It is the time when the receiver successfully receives the packet. Due to fixed 24
simulation time, there may be packets waiting to be completed at the end of a 25
simulation run. For these packets, the delivered time is the end of the simulation. 26
PCTA(m,n,k) = arrival time of the k-th packet call for user(m,n), it is the time when 27
the first packet of the packet call arrives at the transmitter side and is put into a 28
queue. 29
PCTD(m,n,k) = delivered time of the k-th packet call for user(m,n). It is the time 30
when the receiver successfully receives the last packet of the packet call. Due to 31
fixed simulation time, there may be packet calls waiting to be completed at the end 32
of a simulation run. For these packet calls, the delivered time is the end of the 33
simulation. 34
The arrival time of a packet call is the time when the first packet of the packet call arrives 35
at the transmitter side and is put into a queue, and the delivered time of a packet call is the 36
time when the last packet of the packet call is successfully received by the receiver, i.e., 37
3GPP2 C.R1002-A v1.0
126
PCTA(m,n,k) = TA(m,n,k,1) and PCTD(m,n,k) = TA(m,n,k,L(m,n,k)). Due to fixed simulation 1
time, there may be packet calls waiting to be completed at the end of a simulation run. For 2
these packet calls, the delivered time is the end of the simulation. Figure D-1 demonstrates 3
the arrival and delivered times for a packet and a packet call. 4
With the above notation, we can now define various throughputs and delays as follows. 5
Data throughput per sector MT
lknmBM N
n
nmK
k
knmL
l
1m 1
),(
1
),,(
1
),,,(
, (D-1) 6
Averaged delay per sector
M N
n
M N
n
nmK
k
nmK
knmPCTAknmPCTD
1m 1
1m 1
),(
1
),(
),,(),,(
, (D-2) 7
Data throughput for user(m,n) T
lknmBnmK
k
knmL
l
),(
1
),,(
1
),,,(
, (D-3) 8
Packet call throughput for user(m,n)
),(
1
),(
1
),,(
1
),,(),,(
),,,(
nmK
k
nmK
k
knmL
l
knmPCTAknmPCTD
lknmB
, (D-4) 9
Averaged packet delay per sector
M N
n
nmK
k
M N
n
nmK
k
knmL
l
knmL1m 1
),(
1
1m 1
),(
1
),,(
1
),,(
l)k,n,TA(m,-l)k,n,TD(m,
, (D-5) 10
Averaged packet delay for user(m,n)
),(
1
),(
1
),,(
1
),,(
),,,(),,,(
nmK
k
nmK
k
knmL
l
knmL
lknmTAlknmTD
, (D-6) 11
Averaged packet call delay for user(m,n)
),(
),,(),,(
),(
1
nmK
knmPCTAknmPCTDnmK
k
. (D-7) 12
3GPP2 C.R1002-A v1.0
127
Packet call is delivered within a simulation run:
0 T
PCTA(m,n,k) PCTD(m,n,k)
Packet call is not delivered by the end of a simulation run:
0 T
PCTA(m,n,k)PCTD(m,n,k) =T
Last packet is not
delivered yet
Time
Time
Frist packet is delivered Last packet is delivered
First packet arrives
Frist packet is delivered
First packet arrives
T
TA(m,n,k,l) TD(m,n,k,l)
T
TA(m,n,k,l)TD(m,n,k,l) =T
Packet is not delivered yet
Time
Time
Packet is delivered
Packet arrives
Packet arrives
Packet is delivered within a simulation run:
Packet is not delivered by the end of a simulation run:
1
Figure D-1: Description of arrival and delivered time for a packet and a packet call. 2
3
3GPP2 C.R1002-A v1.0
128
APPENDIX E: LINK BUDGET 1
This appendix contains the forward and reverse link budget templates for 1xEV-DV and 2
1xEV-DO evaluation. These templates, when filled out in the evaluation process, provide 3
maximum sustainable path loss and maximum range for different data rates or throughput 4
levels. 5
The forward link templates differ from the typical approach where a fixed data rate is used 6
to determine the range. This is because some system proposals use adaptive modulation 7
and coding schemes to change the data rate on the forward link to take advantage of the 8
changing channel conditions. Three fixed per-sector forward throughput levels (for four 9
statistically identical mobile stations) are used to determine the maximum supportable 10
path loss for data users. The way to fill out this forward link budget for data is: 11
Determine the per-sector forward link throughput to be used in the link budget. This 12
shall be at a level all proposals can achieve. 13
Set the geometry/location of the mobile station (interior or cell edge) and per-sector 14
multi-path fading profile (Pedestrian A, 1-path 3 km/hr, Pedestrian B, 3-path, 10 15
km/hr, Vehicular A, 2-path, 30 km/hr, Pedestrian A, 1-path, 120 km/hr, or Single 16
path Rician). 17
Simulate the link performance and obtain the FER vs Eb/Nt plot for each of the 18
possible data rates the adaptive scheme uses 19
Simulate sector throughput with the set of performance curves for the possible data 20
rates for one mobile station. If applicable, scheduling is simulated, too. With all other 21
link budget parameters fixed, adjust the path loss so that the throughput reaches the 22
specified level above. 23
To account for the additional range the adaptive schemes can achieve, a ―multi-user 24
diversity gain‖ is included. This is obtained by running the sector simulator with 25
scheduler among four mobile stations. These four mobile stations have the same path 26
loss to the base station and independent but statistically identical fading process. 27
When the simulated sector throughput is the same as the single-user case above, the 28
range increase (in dB) over the single-mobile case is the multi-user diversity gain. 29
Note for 1xEV-DV systems: Since the forward link channelization (e.g., Walsh) code space is 30
limited, the forward link overhead channels and control channels have to be accounted for. 31
The available code space is to be specified and justified by the proponents of the DV 32
proposals. The sector throughput simulation above has to be carried out within the 33
available code space for traffic channels. 34
Also, to make the comparison easy, no transmit diversity is to be used in these link budgets 35
on either link. 36
Link-Budget Templates 37
Tables E-1 and E-2 give the proposed link-budget templates for the reverse link and the 38
forward link, respectively. 39
3GPP2 C.R1002-A v1.0
129
Table E-1 Link-Budget Template for the Reverse Link 1
Reverse-Link Item Value Comments
Test Environment
Input (per antenna)
Pedestrian A, 1-path 3 km/hr
Pedestrian B, 3-path, 10 km/hr
Vehicular A, 2-path, 30 km/hr
Pedestrian A, 1-path, 120 km/hr
Single path Rician
Test Service Descriptive Text
Type of Traffic
Input: Voice Only, Data Only, or Data
with Voice at a power division that gets
the same range for 9.6 kbps and the data
rate under consideration.
Chip Rate (Mcps) 1.2288 Fixed Value
Transmitter Power (dBm) 23
Fixed Value: Maximum available total
power after any peak-to-average backoff
has been taken out.1
Fraction of the Power Used
for the Traffic Channel (dB)
Input: Varies by Approach. Traffic
Channel Ec/Ior.
Traffic Channel
Transmitter Power (dBm) Calculated
Cable, Connector, and
Combiner Losses (dB) 0 Fixed Value
Transmitter Antenna Gain (dBi) –1 Fixed Value:
Transmitter EIRP per
Traffic Channel (dBm) Calculated
Receiver Antenna Gain (dBi) 17 Fixed Value
Cable and Connector
Losses (dB) 2 Fixed Value
Receiver Noise Figure (dB) 5 Fixed Value
Thermal Noise Density (dBm/Hz) Calculated: –174 + Receiver Noise Figure
Rise Over Thermal,
(I0 + N0)/N0 (dB) 7 Fixed Value
Noise Plus Interference
Density (dBm/Hz) Calculated
1 In other words, the actual PA has to be larger than 23 dBm since it has to take into account the
peak to average ratio.
3GPP2 C.R1002-A v1.0
130
Reverse-Link Item Value Comments
FER without
Retransmissions
Fixed Value: For voice, it is 0.10 when in
HO and 0.01 when not in HO. For data, it
is 0.20 when in HO and 0.04 when not in
HO.
Data Rate (kbps)
Fixed Value: For voice, it is 9.6. For data,
link budgets should be provided for data
rates of 9.6, 38.4, and the highest
supported rate.
Required per Antenna Traffic Eb/(N0 + I0) with
Two Antennas
(dB)
Input: Varies by Approach. Value with
power control off since the mobile is at
the cell edge. Determined by a link
simulation.
Receiver Sensitivity per
Antenna (dBm) Calculated
Building Penetration Loss
for Outdoor-to-Indoor or
Vehicular Penetration Loss
for Vehicular
(dB)
20 for the Building Penetration Loss and
10 for the Vehicular Penetration Loss.
Log-Normal Fade Margin (dB) Fixed Value: 11.2 for Pedestrian and
Rician channel, 11.4 for Vehicular.
Handoff Gain (dB)
Fixed Value: Isolated cell fade margin –
fade margin with best-of-two cell-edge
coverage. 5.0 for Pedestrian, Rician
channel and Vehicular.
Maximum Path Loss (dB) Calculated
Maximum Range (m) Calculated from the maximum path loss
(see Note 1).
1
2
3GPP2 C.R1002-A v1.0
131
Table E-2 Link-Budget Template for the Forward Link 1
Forward-Link Item Value Comments
Test Environment
Input: Pedestrian A, 1-path 3 km/hr
Pedestrian B, 3-path, 10 km/hr
Vehicular A, 2-path, 30 km/hr
Pedestrian A, 1-path, 120 km/hr
Single path Rician
Mobile Velocity (km/h) Input
Test Service Descriptive Text
Type of Traffic to a
Particular Mobile
Input: Voice Only, Data Only, or Data
with Voice with 50-50 split of non-
overhead power
Mobile Location Cell Edge in HO, Cell Edge Not in HO, or
Cell Interior Not in HO. See Note 3.
Chip Rate (Mcps) 1.2288 Fixed Value
Maximum PA Power (dBm) 43 Fixed Value
Peak-to-Average Backoff at
99.9 percentile for the
used channel
configuration
(dB) Input: Varies by Approach
Total Transmitter Output
Power (dBm) Calculated
Fraction of the Power Used
for Voice Traffic
Total voice Ec/Ior, including power
control subchannel.
For Voice Only, Data Only, and Data
with Voice.
Fraction of the Power Used
for Data Traffic
Total data Ec/Ior.
Fixed value of 0.0 for Voice Only
Specify Data Only and Data with Voice.
Fraction of the Power Used
for the Pilot, Synch, and
Paging Channels
Total overhead Ec/Ior.
Fixed value of 0.2
Fraction of the Power Used
for the Other Control
Channels1
Varies by approaches
1 This and the 3 terms above it sum to 1
3GPP2 C.R1002-A v1.0
132
Forward-Link Item Value Comments
Fraction of the Code Space
Used for Data Traffic
Fixed value of 0.0 for Voice Only.
Input that varies by approach for Data
Only and Data with Voice.
Fraction of the Code Space
Used for the Pilot, Control
Traffic, and Margin
Calculated. The sum of this value and the
previous code space is limited by the total
code space available.
Fraction of the Power Used
for the Link-Budget Traffic
Channel
(dB)
Traffic Ec/Ior.
Input: Varies by Approach. This fraction
is less than or equal to the total power
fraction for voice or data noted above 1.
Fraction of Code Space
Used for the Link-Budget
Traffic Channel
Input: Varies by Approach. This fraction
is less than or equal to the total channel
resource fraction for voice or data noted
above.2
Cable, Connector, and
Combiner Losses (dB) 2 Fixed Value
Transmitter Antenna Gain (dBi) 17 Fixed Value
Traffic Channel EIRP (dBm) Calculated
Total Transmitter EIRP (dBm) Calculated
Receiver Antenna Gain (dBi) –1 Fixed Value
Cable and Connector
Losses (dB) 0 Fixed Value
Receiver Noise Figure (dB) 10 Fixed Value
Thermal Noise Density (dBm/Hz) Calculated: –174 + Receiver Noise Figure
Voice FER Fixed input of 0.01 for voice and N/A for
data.
Voice Data Rate (kbps) Fixed input of 9.6 for Voice Only and N/A
for Data Only and Data with Voice.
1 For example, power control subchannel power consumption has to be excluded.
2 This term does not reduce the amount of power available for the traffic channel but might limit the
sector throughput.
3GPP2 C.R1002-A v1.0
133
Forward-Link Item Value Comments
Sector Data Throughput (kbps)
Input: N/A for Voice Only and at least
three different values for Data Only and
Data with Voice. This is the throughput
of just the data portion of the link. The
throughput is without RLP packet
retransmissions, where packets may
consist of multiple slots and subpackets
that are not transmitted contiguously.
Required Îor/N0 with a
Single User (dB)
Input: Determined by a link simulation
and a network simulation.
Multiple-User Data Îor/N0
Gain (dB)
Input: Gain with four scheduled users
compared to that required for a single
user for the same total sector
throughputs and total (over the 4 users)
average fractional powers. Determined by
a network simulation with the specified
channel conditions.
Receiver Interference
Density, Ioc (dBm/Hz)
Calculated based on the equations in
Note 3.
Noise Plus Interference
Density (dBm/Hz) Calculated
Receiver Sensitivity (dBm) Calculated
Building Penetration Loss
for Outdoor-to-Indoor or
Vehicular Penetration Loss
for Vehicular
(dB) 20 for the Building Penetration Loss and
10 for the Vehicular Penetration Loss.
Log-Normal Fade Margin (dB) Fixed Value: 11.2 for Pedestrian and
Rician channel, 11.4 for Vehicular.
Handoff Gain (dB)
Fixed Value: the values for Pedestrian
Rician channel, and Vehicular
environments are 5.0 for best-of-two HO
and 6.2 for sum-of-two HO1. Use 0 for no
HO gain.
Maximum Path Loss (dB) Calculated
Maximum Range (m) Calculated from the maximum path loss
(see Note 1).
1 Sum of two is the typical soft handoff where multiple sectors transmit to the MS and the MS
combines the traffic channels across sectors. The Best-of-two HO is closer to hard handoff in that
only one of the sectors is transmitting to the MS at any given time.
3GPP2 C.R1002-A v1.0
134
The following are some notes for the link budgets. 1
Note 1: The maximum range in meters, R, is calculated from the maximum path loss in 2
dB, L. In [20], the specified equations depend on the test environment as follows: 3
( 38.462)/20 ( 37)/30
( 38.462)/20 ( 28.031)/40
( 38.462)/20 ( 15.352)/37.6
min(10 ,10 ) for Indoor Office
min(10 ,10 ) for Pedestrian and Outdoor-to-Indoor
min(10 ,10 ) for Vehicular
L L
L L
L L
R
(E-1) 4
In [23], the range for all test environments is 5
( 28.6)/3510 LR (E-2) 6
The range equations from [23] are recommended. 7
Note 2: The propagation index and log-normal fading sigma values specified in [20] are 8
shown in Table E-3. 9
Table E-3 Propagation Index and Log-Normal Sigma Values from [20] 10
Test Environment Propagation
Index
Log-Normal Sigma
(dB)
Indoor Office 3 12.0
Pedestrian 4 10.0
Outdoor-to-Indoor with the Building
Penetration Loss 4 14.4
Vehicular 3.76 10.0
11
Note 3: When the mobile is at the cell/sector edge, assume, as in [24], that it is receiving a 12
signal from three cells/sectors with two of the signals having the same average received 13
power density and the other having 6 dB less. When the mobile is in the cell/sector 14
interior, assume that it is receiving a signal from two cells/sectors with the average received 15
power density from one being 6 dB less. If the signal density received from the target cell 16
(the one where the link budget range is being calculated) is Itc, the other-cell interference 17
density is Ioc, and the total received signal power signal is Îor, then 18
0.25 for the cell edge in handoff (HO)
1.25 for the cell edge not in HO
0.25 for the cell interior not in HO
tc
oc tc
tc
I
I I
I
(E-3) 19
and 20
2 for the cell edge in HOˆ for the cell edge not in HO
for the cell interior not in HO
tc
or tc
tc
I
I I
I
(E-4) 21
Note 4: The frequency shall be 2 GHz, as in [20]. 22
23
3GPP2 C.R1002-A v1.0
135
APPENDIX F: QUASI-STATIC METHOD FOR LINK FRAME ERASURES GENERATION 1
AND DYNAMICALLY SIMULATED FORWARD LINK OVERHEAD CHANNELS 2
The following Appendix describes in more details the quasi-static method for modeling the 3
link level performance of 1xEV-DV PDCH and 1xEV-DO FTC and dynamically simulated 4
forward link overhead channels in the system level. 5
Definitions 6
Aggregate Es/Nt Metric for the AWGN Channel 7
The aggregate Es/Nt, denoted s tE /N is defined as 8
s tE /N 10 s t1
110log .(E /N ) ,
n
j jj
NN
(F-1) 9
where 10
1. N equals the number of information bits (i.e., the encoder packet size). 11
2. jN equals the number of modulation symbols transmitted in slot j. 12
3. n is the number of slots over which the transmission occurs. This includes both the 13
original transmission, and retransmissions, if any. For example, if the duration of 14
the original transmission is 4 slots, and that of the retransmission is 2 slots, then 15
n=6. 16
4. s t(E /N ) , 1,..., ,j j n is the SNR per modulation symbol for slot j. These terms are not 17
in dB. 18
5. Note that s tE /N =Eb/No because N equals the number of information bits. 19
6. s t(E /N ) , 1,..., ,j j n is the Es/Nt observed after Rayleigh (or Jakes) fading. 20
The functional relationship1 between s tE /N and BLER for the base 1/5 turbo code over 21
the AWGN channel with m-ary modulation will be denoted by 5mf . For the sake of 22
convenience, the superscript in 5mf is dropped for QPSK modulation. Thus, 5f denotes the 23
functional relationship between s tE /N and BLER for the base 1/5 turbo code over the 24
AWGN channel, when using QPSK modulation. The following table, which is for illustration 25
purposes only, is an example of what 5f will look like. 26
27
1 This functional relationship will depend on the encoder packet size. So, the proponent will need to
generate this relationship for every encoder packet size that has been defined.
3GPP2 C.R1002-A v1.0
136
s tE /N (in dB) BLER
-0.75 1
-0.55 0.986842
-0.35 0.707547
-0.15 0.205479
0.05 0.015538
0.25 0.00019
Effective Coding Rate 1
.Total number of information bits
Effective coding rateTotal number of information bits +
Total number of unique parity bits received so far
(F-2) 2
Clearly, the effective coding rate remains unchanged in the case of pure Chase combining. 3
In the case of incremental redundancy based schemes, the effective coding rate continues 4
to decrease with every retransmission, until it equals the base turbo coding rate. 5
Prediction Error Rate 6
Difference in the actual and predicted BLER at different average Es/Nt. In particular, one 7
look at the prediction error percentage, defined as 8
(Actual number of block errors Predicted number of block errors).100
Total number of blocks transmitted (F-3) 9
The motivation behind this measure is as follows. The actual throughput seen at a given 10
Es/Nt is 11
Actual number of block errors1 (Peak rate),
Total number of blocks transmitted (F-4) 12
while the predicted throughput is 13
Predicted number of block errors1 (Peak rate).
Total number of blocks transmitted (F-5) 14
Therefore, the difference in actual throughput and predicted throughput is 15
(Actual number of block errors Predicted number of block errors)(Peak rate).
Total number of blocks transmitted(F-6) 16
Normalizing with respect to peak rate, and taking a percentage yields the first expression. 17
Catastrophic Error Rate 18
Number or percentage of ―Catastrophic Errors.‖ For an AWGN channel and a given effective 19
coding rate, the BLER is almost 0 if s tE /N 0T dB, and the BLER is 1 if
s tE /N 1T dB. 20
For turbo codes, 0 1 1T T dB. For the fading channel, one declare a ―catastrophic error‖ if 21
one of the following two events occur 22
3GPP2 C.R1002-A v1.0
137
s tE /N 0T AND the block is actually in error, or 1
s tE /N 1T AND the block is actually NOT in error. 2
The significance of a catastrophic error is as follows. The aggregate Es/Nt metric only 3
captures a first order statistic of the channel variations. If the second order variations of 4
the channel were to play a significant role in determining BLERs, then s tE /N would prove 5
to be insufficient in characterizing BLERs, and this, in turn, would lead to a large number 6
of catastrophic errors. Few catastrophic errors, therefore, imply that the second order 7
statistics of the channel are not important as long as s tE /N 0T or
s tE /N 1T . Since 8
0 1 1T T dB (i.e., small), this would also imply that s tE /N is a sufficient for predicting 9
BLERs. 10
Puncturing Penalty (or Coding Gain) 11
For a code with effective coding rate 1/M, where 1/M > 1/5, and modulation order m, the 12
puncturing penalty is defined to be the additional s tE /N (or Eb/No) (in dB) required (with 13
respect to the base 1/5 turbo code) to achieve a BLER of 0.001 over an AWGN channel. 14
The puncturing penalty1 is denoted by mMC . The superscript here refers to the modulation, 15
while the subscript refers to the effective coding rate. For the sake of convenience, the 16
superscript is dropped in the case of QPSK modulation. The following table, which is for 17
illustration purposes only, shows the puncturing penalty for a few sample cases for an 18
encoder packet size of 3072 bits for QPSK modulation. s tE /N required to achieve a BLER of 19
0.001 for the base turbo code is approximately 0.2 dB. 20
21
Effective coding rate s tE /N required Puncturing penalty in dB
¼ 0.4 0.2
1/3 0.75 0.55
½ 1.6 1.4
This penalty applies both for IR and Chase combining. 22
Doppler Penalty 23
,D M is denoted as the Doppler penalty at Doppler = D Hz, for an effective coding rate of 24
1/M. Simulations show that ,D M =0, if D < 30Hz. For the values of M that were considered, 25
at Dopplers around 100 Hz, the penalty lay between 0.2 and 0.6 dB. This penalty applies 26
both for IR and Chase combining. 27
Demapping Penalty 28
Suppose m-ary (m > 4) modulation is being used for transmission over the channel. 2 cases 29
are considered here. 30
1 Note that MC (expressed in dB) is always positive.
3GPP2 C.R1002-A v1.0
138
1. Pure Chase combining. This implies that the modulation and coding rate does not 1
change for retransmissions, AND the combining of the repeated symbols is done at the 2
modulation symbol level. These combined symbols are then demapped to soft bits 3
(which are then input to the turbo decoder). In this case, 5mf should be used for 4
predicting BLER. Recall that 5mf denotes the functional relationship between 5
s tE /N and BLER for the base 1/5 turbo code over the AWGN channel with m-ary 6
modulation. So, there is no need for explicitly introducing demapping penalties in 7
this case1. If the effective code rate of the transmission is greater than 1/5, coding 8
penalties should also be applied before prediction. 9
2. For all schemes that do not fall in the category 1 above, demapping penalties apply. 10
Therefore this includes 11
Schemes which use IR (pure or otherwise), and 12
Schemes that use Chase combining, BUT the modulation or the coding rate is 13
different for retransmissions, OR the modulation symbols are first demapped 14
to soft bits, and then combined. 15
Demapping penalties are a function of the modulation being used, and the16
s t(E /N ) ,j which denotes the SNR for a modulation symbol in slot j. Precisely,17
s tE /N is given by 18
s ts t
s tE /N 10
1 (E /N )
.(E /N )110log ,
j
n j j
j
N
N
(F-7) 19
where x denotes the demapping penalty when s tE / N ,x where Es/Nt denotes the 20
modulation symbol SNR. Note that the terms s t(E /N ) ,j and x are not in dB, and 21
1.x Simulations indicate that x is a monotone decreasing function of x, and is 22
typically greater than 2. Upon obtaining s tE /N , as defined in F-7, 5f should be 23
used for predicting BLER. Recall that 5f denotes the functional relationship 24
between s tE /N and BLER for the base 1/5 turbo code over the AWGN channel with 25
QPSK modulation. 26
Flowcharts 27
The first flowchart illustrates the method when the QPSK modulation is used. As the figure 28
illustrates, the following values are needed for aggregate Es/Nt method: 29
1. 5f denotes the functional relationship between s tE /N and BLER for the base 1/5 30
turbo code. 31
2. Effective coding rates, and the corresponding puncturing penalties (or coding gains) 32
for various effective rates. These puncturing penalties will, in general, depend on the 33
size of the encoder packet, i.e., the number of information bits. 34
1 In this case, demapping penalties are being implicitly accounted for in 5mf .
3GPP2 C.R1002-A v1.0
139
3. The Doppler penalty for various Dopplers and encoder packet sizes. 1
Figure F-1 Flowchart for QPSK modulation 2
3
The next flowchart illustrates the methodology for higher order modulations when IR is 4
being used. 5
Effective coding rate = 1/M
s tE /N 10 s t
1
1 Calculate 10 log .(E /N ) .
n
j jj
NN
s t s tE /N E /N
Introduce puncturing penalty:
, where
is in dB.
M
M
C
C
s t s tE /N E /N ,
,
Introduce Doppler penalty:
,where
is in dB.
D M
D M
s tE /N 5 Use the adjusted and to
determine whether the block is in error.
f
3GPP2 C.R1002-A v1.0
140
Figure F-2 Prediction methodology for higher order modulations without pure Chase 1
combining 2
3
The next flowchart illustrates the prediction methodology for higher order modulations 4
when pure Chase combining is being used. In particular, m-ary modulation is being used. 5
The puncturing penalty used in step 3 of the flowchart is mMC and not MC , while in the last 6
step of the flowchart the table being used is 5mf and not 5f . This distinction is important; 7
recall that 5f is the relationship between s tE /N and BLER with QPSK modulation, while 8
5mf is the relationship between
s tE /N and BLER with m-ary modulation (m > 4). Similarly, 9
mMC is the puncturing penalty for the rate 1/M code with m-ary modulation. 10
Effective coding rate = 1/M
s t
s t
s tE /N 10
1 (E /N )
.(E /N )1 Calculate 10 log .
j
n j j
j
N
N
s t s tE /N E /N
Introduce puncturing penalty:
, where
is in dB.
M
M
C
C
s t s tE /N E /N ,
,
Introduce Doppler penalty:
,where
is in dB.
D M
D M
s tE /N 5 Use the adjusted and to
determine whether the block is in error.
f
3GPP2 C.R1002-A v1.0
141
Figure F-3 Prediction methodology for higher order modulations with pure Chase 1
combining (this corresponds to Case 1) 2
3
Obtaining Coding Gains or Puncturing Penalties 4
Let s tE /N
Mdenote the
s tE /N required to achieve a BLER of 0.001 over an AWGN 5
channel when the effective coding rate is 1/M. Then, the coding gain or puncturing penalty 6
for the effective coding rate of 1/M is simply 7
s t s tE /N E /N
5.
M (F-8) 8
Puncturing penalties depend on the encoder packet size N, and should be evaluated 9
separately for each encoder packet size in use. 10
In the case of pure Chase combining, puncturing penalties should be obtained for 11
different modulations as well. For cases without pure Chase combining, puncturing 12
penalties should not depend on the modulation in use. However, one can determine 13
them separately for each modulation also. 14
Obtaining Doppler Penalties 15
The Doppler penalty is obtained by simple curve fitting. 16
Effective coding rate = 1/M
s tE /N 10 s t
1
1 Calculate 10 log .(E /N ) .
n
j jj
NN
s t s tE /N E /N
Introduce puncturing penalty:
, where
is in dB.
mM
mM
C
C
s t s tE /N E /N ,
,
Introduce Doppler penalty:
,where
is in dB.
D M
D M
s tE /N 5 Use the adjusted and to determine
whether the block is in error or not.
mf
3GPP2 C.R1002-A v1.0
142
Fix 1/M (the effective coding rate), n (the duration of the transmission, i.e., number of 1
slots), N (the encoder packet size), and the D (the Doppler). For various average geometries 2
(from x dB to y dB, say), one run link level simulations to obtain sample values of s tE /N 3
and an indicator of whether the packet is in error at the given value of s tE /N . 4
For each collected sample pair ( s tE /N , indicator of block error), one applies the predictor, 5
and chooses the value of the Doppler penalty that results in the best predictor 6
performance. Precisely, one first fixes a value of ,D M . Then, for each sample collected, one 7
apply the Doppler penalty, and the puncturing penalty to obtain the modified s tE /N , i.e., 8
s tE /N 10 s t ,
1
110log .(E /N ) .
n
j j D M Mj
N CN
(F-9) 9
one then use 5f and the modified s tE /N above to predict whether the block is in error. For 10
each average geometry, one evaluates the performance of the predictor by computing the 11
prediction error rates and catastrophic error rates. Finally, the value of Doppler penalty, 12
which results in the best prediction performance is chosen to be the value of ,D M . 13
The following flowcharts illustrate this method. 14
Figure F-4 Obtaining sample values of
s tE /N and indicators of packet errors 15
Transmit packet
s tE /N 10 s t
1
1 Calculate 10 log .(E /N ) .
n
j jj
NN
For each value of average geometry from x dB to y dB,
run the following link level simulation. For each value
of average geometry, the recommended simulation
duration is 25000*n.
Note that the average geometry AND the single-path
(Jakes) fading gives us the Es/Nt in each slot.
s tE /N Store and the indicator of
whether the packet was in error.
3GPP2 C.R1002-A v1.0
143
1
Figure F-5 Determining the Doppler penalty by evaluating the predictor performance 2
s t s tE /N E /N ,
For each sample, introduce Doppler
penalty: .D M
s tE /N 5 Use the adjusted and to
predict whether the block is in error.
f
For each value of average geometry apply the
predictor and a Doppler penalty on the samples
collected to evaluate the prediction performance.
In particular, we evaluate the prediction error
and catastrophic error rates for the predictor at
the chosen value of Doppler penalty. Prediction
error and catastrophic errors were defined in an
earlier contribution.
These error rates can be obtained in the following
manner:
Determine if it was a ―catastrophic error.‖ If yes,
then update appropriate counter.
If a packet error is predicted, update appropriate
counter.
These values will be used to obtain the final
catastrophic error and prediction error rates at the
given geometry.
s t s tE /N E /N
Introduce puncturing penalty:
, where
is in dB.
M
M
C
C
3GPP2 C.R1002-A v1.0
144
Doppler penalties depend on the encoder packet size and the effective coding rate. 1
However, to limit the simulation effort, extrapolation techniques may be used to 2
obtain Doppler penalties for different effective coding rates and encoder packet 3
sizes. 4
Obtaining Demapping Penalties 5
Fix N, the encoder packet size, and m, the modulation. Set the Doppler to 10 Hz. Let z dB 6
be the Es/Nt required to achieve a BLER of 0.001 over an AWGN channel for the given 7
encoder packet size and modulation when using the base 1/5 turbo code. Then, run the 8
following link level simulations over the fading channel for average geometries ranging from 9
z-12 dB to z-2 dB1. Transmit the encoder packet; schedule retransmissions, if necessary; 10
continue retransmissions till the encoder packet is successfully decoded. Let n denote the 11
number of transmissions necessary for the packet to be successfully decoded, and 12
( ), 1,..., ,n i i n the number of slots used after the i-th transmission. For the given encoder 13
packet, we collect the following samples: s t(E /N ) , 1,..., ( ).j j n n 14
Next, fix a set of demapping penalties x . Recall that the demapping penalties are a 15
function of the modulation symbol SNR x. (one can assume that the demapping penalties 16
are a piecewise linear function of modulation symbol SNR.) Then, calculate 17
s tE /N
, 1,..., ,i i n the aggregate SNR after applying the demapping penalties and the 18
puncturing penalties, after the i-th transmission as follows: 19
s ts t
( ) s t10E /N
1 (E /N )
.(E /N )110log , 1,..., ,
j
n i j ji iM
j
NC i n
N
(F-10) 20
where iMC is the puncturing penalty after the i-th transmission. Next, for each 1,..., ,i n 21
use 5f and s tE /N
i to predict whether the packet was in error. Note that for each 22
1,..., 1,i n the packet is actually in error. 23
Finally, for each average geometry, evaluate the performance of the predictor by computing 24
the prediction error rates and catastrophic error rates. The demapping penalties, which 25
result in the best prediction performance, are then chosen to be the values of x . 26
Obtaining the demapping penalties require a large number of samples. So, it is 27
recommended that for each average geometry, one attempt at least 25000 28
transmissions of encoder packet transmissions. 29
Demapping penalties should not depend on the encoder packet size. However, one 30
can try to obtain separate demapping penalties for each such size defined. 31
Demapping penalties do not depend on the effective coding rate. 32
1 This range of average geometries may be reduced if it is expected that the higher order modulation
is not going to be used at really low values of Es/Nt.
3GPP2 C.R1002-A v1.0
145
Different modulations will have different demapping penalties. So, demapping 1
penalties need to be determined for each modulation being used. 2
Recommendations 3
Puncturing penalties (or coding gains) result in extremely accurate predictions if 4
1/M < 0.75. If 1/M 0.75, then the use of the actual functional relationship 5
between s tE /N and BLER for the corresponding value of M is recommended. 6
Doppler penalties can be obtained by ―intelligent‖ extrapolation. For example, for 7
D=100Hz, if the Doppler penalty is 0.25, 0.5 dB and 1 dB for 1/M=0.2, 0.5, and 8
0.75, respectively, then suitable extrapolation can be used to obtain Doppler 9
penalties for all values of M in between. 10
11
3GPP2 C.R1002-A v1.0
146
APPENDIX G: EQUALIZATION 1
The following Appendix describes a possible technique to reduce multi-path interference. 2
This technique is optional; however, if equalization is used in a proposal, the technique 3
discussed below shall be used. 4
Equalization (optional) 5
Equalization may be used to reduce the FURP. In the following text, a procedure is provided 6
for evaluating the symbol signal-to-noise ratio of the ideal MMSE equalizer for channels 7
randomly drawn according to the ITU models. The performance of the ideal MMSE 8
equalizer is captured in equation (2) below. A second simulation procedure is described 9
which accounts for the implementation losses associated with the use the MMSE equalizer. 10
The performance of this non-ideal approximation of the MMSE equalizer is captured in 11
equation (14) below. 12
The procedure for the implementation of equalization in the simulation will be the following: 13
i) A random draw is taken from the appropriate ITU channel model (Ped A, Ped B, Veh 14
A). 15
ii) The random draw is convolved with the autocorrelation of the chip filter; 16
iii) Let the complex vector 211
...,,, 1 LLL fff f denote the vector of chip-spaced 17
waveform samples. The vector f is normalized to unit energy. 18
iv) Let the complex vector 433
...,,, 1 LLL ggg g of length 134 LL have coefficients 19
given by 20
else
LLiLLfg
i
i0
},min{},max{ 4231 . (G-1) 21
In general, the index 3L may be chosen to optimize the performance of an equalizer 22
of length 134 LL . Default values of 3L and 4L will be 13 L and 84 L , so 23
that the equalizer has a total of ten taps. 24
v) The symbol signal-to-noise ratio for the equalizer is given by 25
26
gg1
H
or
c
MMSEt
s
I
EN
N
E . (G-2) 27
28
where N is the number of chips per symbol, orc IE is the allocation for the given 29
Walsh code, and ocor II is the geometry. The matrix Ω is given by 30
3GPP2 C.R1002-A v1.0
147
43, ,21
LmlLlmI
Iff
or
oc
lkLkL
klmkml
, (G-3) 1
where 2
else
nn
0
01 . (G-4) 3
4
Note that orI is the intra-cell interference power for the given instantiation of the channel 5
f . It is not the intra-cell power averaged over the distribution of f . 6
The procedure used to account for implementation losses associated with the use of MMSE 7
equalization will be the following: 8
i) Let 211
...,,, 1 LLL fff Δf denote a zero-mean Gaussian random vector with 9
covariance matrix Σ with elements 10
11
lmI
I
I
Eff
I
EM
oc
or
or
pc
lkLk
klmk
or
pc
ml
1
,
10
1
,1, , (G-5) 12
13
where orpc IE , is the pilot allocation. The integer M should be chosen as large as 14
possible subject to the condition 15
Dopplermax
1,228,800
Dopplermax
ratechipM . (G-6) 16
In order to simplify the generation of samples of the random vector Δf , the diagonal 17
covariance matrix Γ with elements 18
.0
,,
else
lmllml (G-7) 19
may be used in place of the true covariance Σ . 20
ii) A random draw is taken of the vector Δf . Let Δfff ˆ . 21
iii) Let x denote a Gaussian random variable with mean 22
ocor IIxE , (G-8) 23
and variance 24
21
var ocor IIN
x . (G-9) 25
3GPP2 C.R1002-A v1.0
148
iv) Let y denote a non-central chi-square random variable with two degrees of freedom, 1
mean 2
ocorl
or
pc
l
or IIfI
E
fNIyE
2
1
,
21
11 , (G-10) 3
and variance 4
.111
2
111
var
2
1
,
2
22
21
,
22
ocorl
or
pc
l
or
ocorl
or
pc
l
IIfI
E
fNI
IIfI
E
fNy
(G-11) 5
6
vi) Random draws are take of x and y . Define 7
.100
0
else
yxyx
y
I
I
oc
or (G-12) 8
vii) Define the matrix Ω with elements given by 9
43
1
10, ,ˆˆˆ LmlLlm
I
Iff
oc
or
lkLk
klmkml
. (G-13) 10
11
viii) The symbol signal-to-noise ratio of the MMSE equalizer based on the estimated 12
channel f and covariance Ω , is given by 13
gg
gg
ˆˆˆˆ
ˆˆ
11
21
H
H
or
c
t
s
I
EN
N
E . (G-14) 14
15
3GPP2 C.R1002-A v1.0
149
APPENDIX H: MAX-LOG-MAP TURBO DECODER METRIC 1
The following Appendix describes the MAX-LOG-MAP turbo decoder metric. 2
Turbo Decoder Metric and Soft Value Generation into Turbo Decoder 3
MAX-LOG-MAP shall be used as turbo decoder metric. 4
The basic block diagram of the mobile station receiver for the simulation is shown in Figure 5
H-1. The M-ary QAM demodulator generates soft decisions as inputs to the Turbo decoder. 6
For M-ary QAM, the soft inputs to the decoder are generated by an approximation1 to the 7
log-likelihood ratio function [3]. First define, 8
22)( )( jjSj
f
i dMindMinKziSji
, 1log,,2,1,0 2 Mi
(H-1)
where M is the modulation alphabet size, i.e. 8, 16, 32 or 64 and 9
nxeAAz j
pd ˆ
ˆ , (H-2)
x is the transmitted QAM symbol, Ad is the traffic channel gain, Ap is the pilot channel gain, 10
je is the complex fading channel gain, and ˆ
ˆ j
p eA is the fading channel estimate 11
obtained from the pilot channel, 12
"0" is ofcomponent : j
th
i yijS , (H-3)
13
"1" is ofcomponent : j
th
i yijS (H-4)
and Kf is a scale factor proportional to the received signal-to-noise ratio. The parameter dj 14
is the Euclidean distance of the received symbol z from the points on the QAM constellation 15
in S or its complement (yj). It is assumed that the value of Ap/Ad is known at the mobile 16
station receiver. In this case the distance metric is computed as shown in Figure H-1 and is 17
written as follows 18
19
222 jpj QzAd iij SSQ or
(H-5)
20
where dA and ˆpA is an estimate formed from the pilot channel after processing 21
through the channel estimation filter as shown in Figure H-1. 22
1 The optimum LLR computation will also include the noise variance due to channel estimation [4].
3GPP2 C.R1002-A v1.0
150
From DCCH
z
+PN
Code
Pilot WalshCodeDespread
TCH WalshCodeDespread
ChannelEstimationFilter
Delay ofCh. Est.Filter
M-ary QAMConstellationVector
2
( )*
DualMinima SoftMetric Generation
Deinterleaver
Decoder
2Pilot Gain
pA
TCH Gain
dA
1
Figure H-1 QAM Receiver Block Diagram 2
3
3GPP2 C.R1002-A v1.0
151
APPENDIX I: 19 CELL WRAP-AROUND IMPLEMENTATION 1
The cell layout is wrap-around to form a toroidal surface to enable faster simulation run 2
times. A toroidal surface is chosen because it can be easily formed from a rhombus by 3
joining the opposing edges. To illustrate the cyclic nature of the wrap-around cell structure, 4
this set of 19 cells is repeated 8 times at rhombus lattice vertices as shown in Figure I-11. 5
Note that the original cell set remains in the center while the 8 sets evenly surround this 6
center set. From the figure, it is clear that by first cutting along the blue lines to obtain a 7
rhombus and then joining the opposing edges of the rhombus can form a toroid. 8
Furthermore, since the toroid is a continuous surface, there are an infinite number of 9
rhombus lattice vertices but only a select few have been shown to illustrate the cyclic 10
nature. 11
1 Note that the set of 19 cells are only repeated for illustrating the cyclic nature of the wrap-around
cell structure. The simulation only contains 19 cells and not 9 sets of 19 cells.
3GPP2 C.R1002-A v1.0
152
10
72
31
98
45
6
18
19
11
12
16
17
15
13
10
72
31
98
45
6
18
19
11
12
16
17
15
14
13
10
72
31
98
45
6
18
19
11
12
16
17
15
14
13
10
72
31
98
45
6
18
19
11
12
16
17
15
14
13
10
72
31
98
45
6
18
19
11
12
16
17
15
14
13
10
72
31
98
45
6
18
19
11
12
16
17
15
14
13
10
72
31
98
45
6
18
19
11
12
16
17
15
14
13
10
72
31
98
45
6
18
19
11
12
16
17
15
14
13
10
72
31
98
45
6
18
19
11
12
16
17
15
14
13
14
10
7
23
1
9
84
56
1819
1112
1617
15
1310
7
23
1
9
84
56
1819
1112
1617
15
14
13
10
7
23
1
9
84
56
1819
1112
1617
15
14
1310
7
23
1
9
84
56
1819
1112
1617
15
14
13
10
7
23
1
9
84
56
1819
1112
1617
15
14
13
10
7
23
1
9
84
56
1819
1112
1617
15
14
13
10
7
23
1
9
84
56
1819
1112
1617
15
14
13
10
7
23
1
9
84
56
1819
1112
1617
15
14
13
10
7
23
1
9
84
56
18
19
1112
1617
15
14
13
14
1
Figure I-1 Wrap-around with ’9’ sets of 19 cells showing the toroidal nature of the 2
wrap-around surface. 3
The antenna orientations to be used in the simulation are defined in Figure 3.3.1.21-2. For 4
simplicity, the clusters in blue from Figure 3.3.1.21-1 have been deleted in this Figure. The 5
distance from any MS to any base station can be obtained from the following algorithm: 6
Define a coordinate system such that the center of cell 1 is at (0,0). The path distance and 7
angle used to compute the path loss and antenna gain of a MS at (x,y) to a BS at (a,b) is the 8
minimum of the following: 9
a. Distance between (x,y) and (a,b); 10
3GPP2 C.R1002-A v1.0
153
b. Distance between (x,y) and );2/38,3( RbRa 1
c. Distance between (x,y) and );2/38,3( RbRa 2
d. Distance between (x,y) and );2/37,5.4( RbRa 3
e. Distance between (x,y) and );2/37,5.4( RbRa 4
f. Distance between (x,y) and );2/3,5.7( RbRa 5
g. Distance between (x,y) and ( 7.5 , 3 / 2)a R b R , 6
where R is the radius of a circle that connects the six vertices of the hexagon. 7
3GPP2 C.R1002-A v1.0
154
12
3
4
13
14
5
1
11
10
2
9
8
7
6
19
18
17
15
16
12
3
4
13
14
5
1
11
10
2
9
8
7
6
19
18
17
15
16
12
3
4
13
14
5
1
11
10
2
9
8
7
6
19
18
17
15
16
12
3
4
13
14
5
1
11
10
2
9
8
7
6
19
18
17
15
16
12
3
4
13
14
5
1
11
10
2
9
8
7
6
19
18
17
15
16
12
3
4
13
14
5
1
11
10
2
9
8
7
6
19
18
17
15
16
12
3
4
13
14
5
1
11
10
2
9
8
7
6
19
18
17
15
16
1
2
Figure I-2: The antenna orientations to be used in the wrap-around simulation. The 3
arrows in the Figure show the directions that the antennas are pointing. 4
5
3GPP2 C.R1002-A v1.0
155
APPENDIX J: LINK LEVEL SIMULATION PARAMETERS 1
Table J-1 and Table J-2 summarize the default simulation parameters to use in generating 2
the reference curves for system level evaluations. 3
Table J-1 Link Level Simulation Parameters for Forward Link 4
Parameter Value
Chip rate 1.2288 Mcps
Carrier frequency 2000 MHz
Channel models Table 2.2.2-1, Table 2.2.2-2,
Table 2.2.2-3, and Table 2.2.2-4
specified in section 2.2.2.
MS Receive diversity 1 RX antenna (2 Uncorrelated
RX antennas are simulated for
1x EV-DO BCMCS)
PC step size 0.5 dB
PC command error rate 4%
PC outer loop 0.5 dB for up-step
PC delay 1 PCG/slot
Channel
estimation for
traffic
channel
Circuit
switched and
packet
switched data
systems (e.g.,
1xEV-DV)
1st order filtering of forward
common pilot channel
Packet
switched data
systems (e.g.,
1xEV-DO,
CDM in
1xEV-DO
BCMCS)
Based on two consecutive pilot
bursts (96 chips each) with
linear interpolation aligned to
achieve zero group delay with
respect to the traffic
Other
Broadcast
Proposals
To be specified by the system
proponents. Channel estimation
imperfections shall be
characterized, and its impact
shall be modeled in system
simulations
3GPP2 C.R1002-A v1.0
156
Power
estimation for
power control
Circuit
switched and
packet
switched data
systems (e.g.,
1xEV-DV)
1 PCG accumulation of forward
PCB subchannel
Packet
switched data
systems (e.g.,
1xEV-DO)
N/A
1
3GPP2 C.R1002-A v1.0
157
Table J-2 Link Level Simulation Parameters for Reverse Link 1
Parameter Value
Chip rate 1.2288 Mcps
Carrier frequency 2000 MHz
Channel models Table 2.2.2-1 and Table 2.2.2-2
specified in section 2.2.2.
BS Receive diversity 2 RX antennas
PC step size 1 dB
PC command error rate 4%
PC outer loop 0.5 dB for up-step
PC delay 1 PCG/slot
Channel
estimation for
traffic
channel
Circuit
switched and
packet
switched data
systems (e.g.,
1xEV-DV)
Equal tap FIR filter (Length 2
PCG) of Reverse pilot channel
Packet
switched data
systems (e.g.,
1xEV-DO)
Equal tap FIR (Length 1 slot) of
Reverse pilot channel
Power
estimation for
power control
Circuit
switched and
packet
switched data
systems (e.g.,
1xEV-DV)
1 PCG accumulation of Reverse
pilot channel
Packet
switched data
systems (e.g.,
1xEV-DO)
N slot(s) accumulation of
Reverse pilot channel Ec/Nt for
a power control update rate of
600/N Hz
2
The link level simulation for reverse link could be performed with one sample per symbol to 3
expedite the simulations. In this case, the multi-path interference shall be modeled as 4
AWGN. The total variance of noise and multi-path interference component in the j-th path 5
of the a-th antenna for the k-th received symbol in the n-th PCG shall be1: 6
1 The Gaussian approximation was considered reasonable tradeoff for simulation simplicity.
3GPP2 C.R1002-A v1.0
158
o
jii
totalciaknjakn
t NEN
_
2),,,(),,,( , (J-1) 1
where totalcE _ is the total transmitted chip energy of all channels. ),,,( jakn is the attenuation 2
factor in the j-th path of the a-th antenna for the k-th received coded symbol in the n-th 3
PCG. 4
5
3GPP2 C.R1002-A v1.0
159
APPENDIX K: JOINT TECHNICAL COMMITTEE (JTC) FADER 1
The JTC fader is best described as a filter bank, as depicted in Figure K-1. Each path of 2
each antenna must be generated by independently. The fade multiplier output values 3
change at the rate 64fd, where fd denotes the Doppler frequency. The output power of the 4
JTC fader is normalized to unity. 5
6
Figure K-1: I and Q Fade Multiplier Generation 7
A good uniform random generator must be used, and the filter clocked several times to “warm-up” before using 8
the first fade sample. 9
Table K-1: Coefficients of the 6-tap FIR Filter 10
-0.040357044
0.086013602
0.454477919
0.454477919
0.086013602
-0.040357044
11
Table K-2: Coefficients of the 8-tap FIR Filter 12
-0.020403315
3GPP2 C.R1002-A v1.0
160
-0.027828149
0.137993831
0.410212183
0.410212183
0.137993831
-0.027828149
-0.020403315
1
Table K-3: Coefficients of the 11-tap Filter 2
0.003612442
-0.012663859
-0.043131662
0.041846468
0.289519221
0.440818845
0.289519221
0.041846468
-0.043131662
-0.012663859
0.003612442
3
Table K-4: Coefficients of the 28-tap Filter 4
0.000134449
-0.001035837
-0.001266781
0.003144467
0.005446165
-0.005705115
-0.015697801
0.005592858
3GPP2 C.R1002-A v1.0
161
0.035808300
0.004331282
-0.073120692
-0.044582508
0.174645729
0.412153786
0.412153786
0.174645729
-0.044582508
-0.073120692
0.004331282
0.035808300
0.005592858
-0.015697801
-0.005705115
0.005446165
0.003144467
-0.001266781
-0.001035837
0.000134449
1
Table K-5: Jakes (Classic) Spectrum IIR Filter Coefficients 2
Denominator Coefficients Numerator Coefficients
1.0000000000000000000 6.5248059900135200e-02
-1.2584602815172037e+01 -5.6908289014580038e-01
8.3781249094641240e+01 2.7480451166883220e+00
-3.8798703729842964e+02 -9.4773135180288293e+00
1.3927662726637102e+03 2.5786482996126544e+01
-4.1039030305379210e+03 -5.8241097311312117e+01
1.0278517997545167e+04 1.1247173657687033e+02
-2.2393748634049065e+04 -1.8904842233132774e+02
3GPP2 C.R1002-A v1.0
162
Denominator Coefficients Numerator Coefficients
1.0000000000000000000 6.5248059900135200e-02
-1.2584602815172037e+01 -5.6908289014580038e-01
8.3781249094641240e+01 2.7480451166883220e+00
4.3133809439790406e+04 2.7936237305345003e+02
-7.4319282567554124e+04 -3.6418631194112885e+02
1.1554604041649372e+05 4.1715604202981109e+02
-1.6315680006218722e+05 -4.1320604132753033e+02
2.1026268214607492e+05 3.3901659663025242e+02
-2.4818342600838441e+05 -2.0059287960205506e+02
2.6898038693500403e+05 2.3734545818966293e+01
-2.6809721585952450e+05 1.5363912802007360e+02
2.4593366073473063e+05 -2.9424154728837402e+02
-2.0763108908648306e+05 3.7359596060374486e+02
1.6120527209223103e+05 -3.8642988435890055e+02
-1.1492103434104947e+05 3.4521505714177903e+02
7.5041686769138993e+04 -2.7265055759799253e+02
-4.4731841330872761e+04 1.9230535924562764e+02
2.4231115205405174e+04 -1.2153980630698008e+02
-1.1857508216082340e+04 6.8773930574859179e+01
5.2013837692697152e+03 -3.4696126060493945e+01
-2.0246855591971096e+03 1.5489134454590417e+01
6.9005516614518956e+02 -6.0495383196143626e+00
-2.0220131802145625e+02 2.0332679679817174e+00
4.9649188538197400e+01 -5.7404157101686004e-01
-9.8333304002079363e+00 1.3121847123296254e-01
1.4770279039919996e+00 -2.2867487042024594e-02
-1.5005452926258436e-01 2.7118486134987282e-03
7.7628588864503741e-03 -1.6371291227220021e-04
1
3GPP2 C.R1002-A v1.0
163
APPENDIX L: LARGEST EXTREME VALUE DISTRIBUTION 1
The Largest Extreme Value distribution is characterized by its probability density function 2
(pdf) and cumulative distribution function (cdf) as follows: 3
0,1
)(
beeb
xfb
ax
eb
ax
(L-1) 4
0,1)(
bexFb
ax
e (L-2) 5
with the moments 6
a b (L-3) 7
2 2 21
6b (L-4) 8
9
where 57722.0 is Euler‘s constant. 10
The random variable with an Extreme Value distribution, X , can be obtained from a 11
random variable with a uniform distribution, Y , with the following transformation: 12
YbaX 1lnln . (L-5) 13
14
3GPP2 C.R1002-A v1.0
164
APPENDIX M: REVERSE LINK OUTPUT MATRICES 1
M.1 Output Matrix 2
M.1.1 1xEV-DV Systems 3
Table M-1 Required statistics output in excel spread sheet for the base station side 4
Spreadsheet
column ID
Name Definition and comments.
1 Simulation run ID Each simulation run is a new drop of
users and a given simulation period.
2 MS ID Mobile station identification for each
simulation run.
3 Traffic type HTTP =1, WAP =2, FTP = 3.
4 Channel model Channel model is A, B, C, D, and E as
defined in Table 2.2.2-1 of Section 2.2.2.
5,6 MS location (km) Mobile station location, two columns (x,
y) relative to the center of the center cell.
7 SHO Geometry (dB) Forward link geometry for the active
sectors. The SHO geometry is
(Ior1+Ior2+…)/(No+Ioc), and does not
have any maximum C/I limitation.
8 Geometry for primary
sector (dB)
Forward link geometry for the primary
(best serving) sector. This geometry is
computed as Ior1/(No+Ioc), and does
not have any maximum C/I limitation.
Note that this is not the geometry used
for generating the ESCAM delay on the
F-PDCH in 1xEV-DV systems.
9 Number of active sectors Number of sectors in the active set.
10,11,12 Active sector index The indices of the serving sectors, best
serving, 2nd best (if any) and 3rd best (if
any) sector, are listed in the 3 respective
columns.
13,14,15 Distance from MS (km) Distance from MS to the center of each
active sector. The considered sectors are
the best serving, 2nd best (if any) and 3rd
best (if any) sector, and are listed in the
3 respective columns.
3GPP2 C.R1002-A v1.0
165
Spreadsheet
column ID
Name Definition and comments.
16,17,18 FL pilot Ec/Io (dB) Forward Link pilot Ec/Io for the best
serving, 2nd best (if any) and 3rd best (if
any) sector. They are listed in the 3
respective columns. This value does not
have any maximum C/I limitation.
19,20,21 Average link gain (dB) Average link gain for the best serving,
2nd best (if any) and 3rd best (if any)
sector. They are listed in the 3
respective columns. Average link gain is
the average propagation loss including
distance dependent path loss, shadow
fade, antenna gains and cable,
connector and other losses.
22 MS average transmit
power (dBm)
MS average transmit power per power
control group (PCG), in dBm.
23 Power outage rate (%) Ratio of the number of PCGs/slots MS
doesn‘t follow the power control
command to the total number of PCGs
in simulation duration (expressed in
percentage).
24, 25, 26 Average combined pilot
received Ec/Nt (dB)
Combined RL pilot received Ec/Nt for
the best serving, 2nd best (if any) and 3rd
best (if any) sector. They are listed in the
3 respective columns.
If sectors are in softer-handoff, then the
combined value should be entered in the
columns corresponding to those sectors.
27, 28, 29 Normalized average
combined pilot received
Ec/Nt (dB)
Normalized average combined pilot
received Ec/Nt for the best serving, 2nd
best (if any) and 3rd best (if any) sector.
They are listed in the 3 respective
columns.
Ec/Nt is normalized by the Pilot
Reference Level of the R-SCH relative to
the Pilot Reference Level of the R-FCH.
If sectors are in softer-handoff, then the
combined value should be entered in the
columns corresponding to those sectors.
30,31,32 Average combined received Average combined R-FCH received traffic
3GPP2 C.R1002-A v1.0
166
Spreadsheet
column ID
Name Definition and comments.
R-FCH Ec/Nt(dB) Ec/Nt for the best serving, 2nd best (if
any) and 3rd best (if any) sector. This
value is defined as accumulated R-FCH
received traffic Ec/Nt divided by
simulation time. Frames during which
SCRM/SCRMM are sent are excluded
from both the numerator and the
denominator. They are listed in the 3
respective columns.
If sectors are in softer-handoff, then the
combined value should be entered in the
columns corresponding to those sectors.
33, 34, 35 Average combined received
R-SCH Ec/Nt(dB)
Average combined R-SCH received
traffic Ec/Nt for the best serving, 2nd
best (if any) and 3rd best (if any) sector.
They are listed in the 3 respective
columns.
If sectors are in softer-handoff, then the
combined value should be entered in the
columns corresponding to those sectors.
36, 37, 38, 39,
40, 41, 42
R-SCH data rate
percentage (kbps)
Percentage of R-SCH data transmitted
with the following rates: 0, 9.6, 19.2,
38.4, 76.8, 153.6, 307.2 [kbps]. Listed
in seven columns. DTX and no-
transmission are treated as rate 0 kbps.
43, 44, 45, 46,
47, 48
R-SCH FER (kbps) R-SCH frame error rate when data is
transmitted with the following rates:
9.6, 19.2, 38.4, 76.8, 153.6, 307.2
[kbps]. Listed in the six respective
columns.
49 R-SCH probability of DTX Probability of a MS being unable to
transmit at the assigned rate either
because it is in power outage or it
doesn‘t have data. It is equal to (# of
DTXed frames on R-SCH) / (# of R-SCH
frames during the duration of the
simulation when assigned rate at MS
side is not 0 kbps)
Frames that are zero rate because the
ESCAM was not received or received late
3GPP2 C.R1002-A v1.0
167
Spreadsheet
column ID
Name Definition and comments.
are not counted.
50 R-SCH average requested
data rate [kbps]
Average R-SCH requested data rate
measured at the MS, in kbps. Does not
exclude CRC and tail bits.
51 R-SCH average assigned
data rate (BTS) [kbps]
Average assigned R-SCH rate measured
at BTS, in kbps. Does not exclude CRC
and tail bits. Does not take into account
whether this rate will be executed or not
by the MS.
52 R-SCH average assigned
data rate (MS) [kbps]
Average assigned R-SCH rate measured
at MS, in kbps. Does not exclude CRC
and tail bits. Differs from 51 in that
assignment message may be lost by the
physical layer or arrive too late.
53 R-SCH average
transmitted data rate
R-SCH average transmitted data rate
over the simulation time (column 63).
Does not exclude CRC and tail bits.
54, 55, 56, 57 Percent of R-FCH data rate
(kbps)
Ratio of the number of 9.6, 4.8, 2.7, 1.5
kbps R-FCH frames carrying data traffic
(Excludes the frames for SCRM and the
interrupted frames due to SCRMM) to
the total number of 20ms frames in the
simulation time. Listed in the 4
respective columns.
58 R-FCH FER The frame error rate of R-FCH carrying
data traffic. It is the ratio of the number
of erased frames to the number of
frames carrying data traffic (Excludes
the frames for SCRM and the
interrupted frames due to SCRMM).
59 R-FCH average
transmitted data rate
(kbps)
Average R-FCH transmitted data rate. It
is the total transmitted bits in R-FCH
frames carrying data traffic (Excludes
the frames for SCRM and the
interrupted frames due to SCRMM)
divided by the simulation time. Does not
exclude CRC and tail bits.
60 Total average transmitted
data rate (kbps)
Sum of the average transmitted data
rates of R-FCH and R-SCH. Does not
3GPP2 C.R1002-A v1.0
168
Spreadsheet
column ID
Name Definition and comments.
exclude CRC and tail bits.
61 MS arrival time (s) The arrival time of each MS (excluding
the setup time).
62 MS departure time (s) The departure time of each MS.
63 Simulation time (s) Simulation time is the period between
the arrival time and the departure time,
excluding the setup time.
64 Total correctly received
bits (kbits)
Total number of correctly received
information bits per user on R-SCH and
R-FCH combined. It excludes the
physical layer overhead (CRC and tail
bits).
65 Average physical layer
data throughput (kbps)
User‘s total number of correctly received
bits (column 64) divided by its
simulation time, as defined in column
63. This throughput does not exclude
zero-padding bits.
66 Total correctly received
bits w overhead (kbits)
Total number of correctly received
information bits per user on R-SCH and
R-FCH combined. It includes the
physical layer overhead (CRC and tail
bits are not removed).
67 Average physical layer
throughput w overhead
(kbps)
User‘s total number of correctly received
bits (column 66) divided by its
simulation time, as defined in column
63.
68 Average packet call
throughput per user
(kbps)
Packet call for FTP, HTTP, WAP and
Gaming users as defined in section 4.2.
Average packet call throughput per user
is defined in section 4.2.1.1.
69 Average packet delay per
user (s)
Average packet delay per user is defined
in Section 3.3.2.2, item 5. Note that
packet time excludes the setup time.
70 Average TCP ACK delay (s) Average TCP ACK delay is defined
Section 3.3.2.2, item 4
71 Average packet call delay
per user (s)
Average packet call delay per user is
defined in Section 3.3.2.2, item 6. Note
that packet call time excludes the setup
3GPP2 C.R1002-A v1.0
169
Spreadsheet
column ID
Name Definition and comments.
time.
72 Correctly received bits on
R-FCH channel (kbits)
Number of correctly received data bits
on R-FCH. This does not include the
CRC and tail bits.
73 Correctly received bits on
R-SCH channel (kbits)
Number of correctly received data bits
on R-SCH. This does not include the
CRC and tail bits.
74 Average R-SCH FER R-SCH FER averaged over all received
frames.
75 Total transmitted R-SCH
frames
Number of total transmitted R-SCH
frames in each simulation run.
76 Total received R-SCH
frames
Number of total received R-SCH frames
in each simulation run.
77 Fraction of ESCAM
messages arriving too late
Total number of ESCAMs received by
MS after action time divided by total
number of ESCAMs received by MS.
78 Percent of R-FCH for
SCRM or SCRMM
Ratio of the number of R-FCH frames
carrying SCRM or SCRMM to the total
number of 20ms frames in the
simulation time. The summation of
columns 54, 55, 56, 57, and 77 should
be one.
79 Total transmitted SCRM or
SCRMM
Number of total transmitted SCRM or
SCRMM in each simulation run.
80 SCRM or SCRMM error
rate
The frame error rate of R-FCH carrying
SCRM or SCRMM.
81, 82, 83 Average combined received
R-CQICH Ec/Nt(dB)
Average combined R-CQICH received
Ec/Nt for the best serving, 2nd best (if
any) and 3rd best (if any) sector. They
are listed in the 3 respective columns.
If sectors are in softer-handoff, then the
combined value should be entered in the
columns corresponding to those sectors.
84 FTP file size [kbytes] FTP upload file size.
85 Average queue size in MS
[kbytes]
Averaged queue size in MS during the
simulation time. This value should be
updated every PCG.
3GPP2 C.R1002-A v1.0
170
Spreadsheet
column ID
Name Definition and comments.
86 Average estimated queue
size in BSC [kbytes]
Averaged estimated queue size in BSC
during the simulation time. This value
should be updated every PCG.
87 Average RLP layer
throughput (kbit/s)
The average RLP layer throughput is
computed as the ratio of the correctly
received bits, excluding the physical
layer CRC bits, tail bits, and any zero-
padding bits, and the simulation time.
88 Gaming packet drop rate The ratio between the total number of
gaming packets dropped because they
stay in MS buffer longer than 160ms,
and the total number of gaming packets
generated by the mobile station
89 Gaming outage indicator ―1‖ if this user in outage, ―0‖ otherwise
1
Table M-2 Required statistics output in excel spread sheet for the base station side 2
Spreadsheet
column ID
Name Definition and comments.
1 Simulation run ID Each simulation run is a new drop of
users and a given simulation period.
2 Sector ID Base Station Sector ID.
3 Average ROT(dB) Average rise over thermal.
4 Prob(ROT_1.25ms > 7 dB) Probability of short term (1.25ms) ROT
exceeding 7 dB.
5 Average load The average load of the sector of
interest, where the load is computed
each power control group (PCG) using
the combined SINR per PCG per
antenna. SINR used is pilot–weighted
combined SINR computed over different
fingers and both antennas of the sector
of interest.
Note that this load is not the estimated
load that the scheduler uses. It is the
actual load.
6 Number of voice MSs Number of voice mobile stations that
3GPP2 C.R1002-A v1.0
171
Spreadsheet
column ID
Name Definition and comments.
have the sector of interest as the
primary sector.
7 Number of voice outage
MSs
Number of voice mobile stations that
have the sector of interest as the
primary sector, and are in outage.
8 Number of data MSs Number of data mobile stations that
have the sector of interest as the
primary sector.
9 Number of data outage
MSs
Number of data mobile stations that
have the sector of interest as the
primary sector, and are in outage.
M.1.2 1xEV-DO Systems 1
Table M-3 Required statistics output in excel spread sheet for the base station side 2
Spreadsheet
column ID
Name Definition and comments.
1 Simulation run ID Each simulation run is a new drop of
users and a given simulation period.
2 MS ID Mobile station identification for each
simulation run.
3 Traffic type HTTP =1, WAP =2, FTP = 3.
4 Channel model Channel model is A, B, C, D, and E as
defined in Table 2.2.2-1 of Section 2.2.2.
5,6 MS location (km) Mobile station location, two columns (x,
y) relative to the center of the center cell.
7 SHO Geometry (dB) Forward link geometry for the active
sectors. The SHO geometry is
(Ior1+Ior2+…)/(No+Ioc), and does not
have any maximum C/I limitation.
8 Geometry for primary
sector (dB)
Forward link geometry for the primary
(best serving) sector. This geometry is
computed as Ior1/(No+Ioc), and does
not have any maximum C/I limitation.
9 Number of active sectors Number of sectors in the active set.
10,11,12 Active sector index The indices of the serving sectors, best
3GPP2 C.R1002-A v1.0
172
Spreadsheet
column ID
Name Definition and comments.
serving, 2nd best (if any) and 3rd best (if
any) sector, are listed in the 3 respective
columns.
13,14,15 Distance from MS (km) Distance from MS to the center of each
active sector. The considered sectors are
the best serving, 2nd best (if any) and 3rd
best (if any) sector, and are listed in the
3 respective columns.
16,17,18 FL pilot Ec/Io (dB) Forward Link pilot Ec/Io for the best
serving, 2nd best (if any) and 3rd best (if
any) sector. They are listed in the 3
respective columns. This value does not
have any maximum C/I limitation.
19,20,21 Average link gain (dB) Average link gain for the best serving,
2nd best (if any) and 3rd best (if any)
sector. They are listed in the 3
respective columns. Average link gain is
the average propagation loss including
distance dependent path loss, shadow
fade, antenna gains and cable,
connector and other losses.
22 MS average transmit
power (dBm)
MS average transmit power per slot, in
dBm.
23 Power outage rate (%) Ratio of the number of slots MS doesn‘t
follow the power control command to
the total number of slots in simulation
duration (expressed in percentage).
24, 25, 26 Average combined pilot
received Ec/Nt (dB)
Combined RL pilot received Ec/Nt for
the best serving, 2nd best (if any) and 3rd
best (if any) sector. They are listed in the
3 respective columns.
If sectors are in softer-handoff, then the
combined value should be entered in the
columns corresponding to those sectors.
27,28,29 Average combined received
RTC Ec/Nt(dB)
Average combined RTC received traffic
Ec/Nt for the best serving, 2nd best (if
any) and 3rd best (if any) sector. They
are listed in the 3 respective columns.
If sectors are in softer-handoff, then the
3GPP2 C.R1002-A v1.0
173
Spreadsheet
column ID
Name Definition and comments.
combined value should be entered in the
columns corresponding to those sectors.
30, 31, 32, 33,
34, 35
RTC data rate percentage
(kbps)
Percentage of RTC data transmitted with
the following rates: 0, 9.6, 19.2, 38.4,
76.8, 153.6[kbps]. Listed in six
columns. More columns shall be added
for any additional set of data rates.
36, 37, 38, 39, 40 RTC FER (kbps) RTC frame error rate when data is
transmitted with the following rates:
9.6, 19.2, 38.4, 76.8, 153.6 [kbps].
Listed in the five respective columns.
More columns shall be added for any
additional set of data rates.
41 RTC average requested
data rate [kbps]
Average RTC requested data rate
measured at the MS, in kbps. Does not
exclude CRC and tail bits. Applicable
only to proposals with centralized RL
scheduler.
42 RTC average assigned data
rate (BTS) [kbps]
Average assigned RTC rate measured at
BTS, in kbps. Does not exclude CRC
and tail bits. Does not take into account
whether this rate will be executed or not
by the MS. Applicable only to proposals
with centralized RL scheduler.
43 RTC average assigned data
rate (MS) [kbps]
Average assigned RTC rate measured at
MS, in kbps. Does not exclude CRC and
tail bits. Differs from 42 in that
assignment message may be lost by the
physical layer or arrive too late.
Applicable only to proposals with
centralized RL scheduler.
44 RTC average transmitted
data rate
RTC average transmitted data rate over
the simulation time (column 47). Does
not exclude CRC and tail bits.
45 MS arrival time (s) The arrival time of each MS (excluding
the setup time).
46 MS departure time (s) The departure time of each MS.
47 Simulation time (s) Simulation time is the period between
the arrival time and the departure time,
3GPP2 C.R1002-A v1.0
174
Spreadsheet
column ID
Name Definition and comments.
excluding the setup time.
48 Total correctly received
bits (kbits)
Total number of correctly received
information bits per user on RTC. It
excludes the physical layer overhead
(CRC and tail bits).
49 Average physical layer
data throughput (kbps)
User‘s total number of correctly received
bits (column 48) divided by its
simulation time, as defined in column
47. This throughput does not exclude
zero-padding bits.
50 Total correctly received
bits w overhead (kbits)
Total number of correctly received
information bits per user on RTC. It
includes the physical layer overhead
(CRC and tail bits are not removed).
51 Average physical layer
throughput w overhead
(kbps)
User‘s total number of correctly received
bits (column 50) divided by its
simulation time, as defined in column
47.
52 Average packet call
throughput per user
(kbps)
Packet call for FTP, HTTP, WAP and
Gaming users as defined in section 4.2.
Average packet call throughput per user
is defined in section 4.2.1.1.
53 Average packet delay per
user (s)
Average packet delay per user is defined
in Section 3.3.2.2, item 5. Note that
packet time excludes the setup time.
54 Average TCP ACK delay (s) Average TCP ACK delay is defined
Section 3.3.2.2, item 4
55 Average packet call delay
per user (s)
Average packet call delay per user is
defined in Section 3.3.2.2, item 6. Note
that packet call time excludes the setup
time.
56 Average RTC PER RTC PER averaged over all received
physical-layer packets.
57 Total transmitted RTC
physical-layer packets
Number of total transmitted RTC
physical-layer packets in each
simulation run.
58 Total received RTC
physical-layer packets
Number of total received RTC physical-
layer packets in each simulation run.
3GPP2 C.R1002-A v1.0
175
Spreadsheet
column ID
Name Definition and comments.
59, 60, 61 Average combined received
DRC (or equivalent)
Ec/Nt(dB)
Average combined DRC (or equivalent)
received Ec/Nt for the best serving, 2nd
best (if any) and 3rd best (if any) sector.
They are listed in the 3 respective
columns.
If sectors are in softer-handoff, then the
combined value should be entered in the
columns corresponding to those sectors.
62 FTP file size [kbytes] FTP upload file size.
63 Average queue size in MS
[kbytes]
Averaged queue size in MS during the
simulation time. This value should be
updated every slot.
64 Average estimated queue
size in BSC [kbytes]
Averaged estimated queue size in BSC
during the simulation time. This value
should be updated every slot.
65 Average RLP layer
throughput (kbit/s)
The average RLP layer throughput is
computed as the ratio of the correctly
received bits, excluding the physical
layer CRC bits, tail bits, and any zero-
padding bits, and the simulation time.
66 Gaming packet drop rate The ratio between the total number of
gaming packets dropped because they
stay in MS buffer longer than 160ms,
and the total number of gaming packets
generated by the mobile station
67 Gaming outage indicator ―1‖ if this user in outage, ―0‖ otherwise
Table M-4 Required statistics output in excel spread sheet for the base station side 1
Spreadsheet
column ID
Name Definition and comments.
1 Simulation run ID Each simulation run is a new drop of
users and a given simulation period.
2 Sector ID Base Station Sector ID.
3 Average ROT(dB) Average rise over thermal.
4 Prob(ROT_1.67 ms > 7 dB) Probability of short term (1.67 ms) ROT
exceeding 7 dB.
5 Average load The average load of the sector of
3GPP2 C.R1002-A v1.0
176
Spreadsheet
column ID
Name Definition and comments.
interest, where the load is computed
each power control group (slot) using the
combined SINR per slot per antenna for
all users in the system. SINR used is
pilot–weighted combined SINR computed
over different fingers and both antennas
of the sector of interest.
Note that this load is not the estimated
load that the scheduler uses. It is the
actual load considering all users in the
system.
6 Average DV load The average load of the sector of
interest, where the load is computed
each power control group (slot) using the
combined SINR per slot per antenna for
all users who consider the sector as a
part of the active set. SINR used is pilot–
weighted combined SINR computed over
different fingers and both antennas of
the sector of interest.
Note that this load is not the estimated
load that the scheduler uses. It is the
actual load considering all users who
considers the sector as a part of the
active set.
6 Number of data MSs Number of data mobile stations that
have the sector of interest as the
primary sector.
7 Number of data outage
MSs
Number of data mobile stations that
have the sector of interest as the
primary sector, and are in outage.
M.2 Definitions 1
Average packet call throughput per user is computed as the ratio of the number of 2
correctly received bits1 (that exclude CRC, tail bits and zero padding bits) during the 3
simulation and the sum of packet call delays during the simulation. 4
1 The correctly received bits are considered and not the correctly received TCP segments because RLP
is not modeled and therefore TCP segment loss rate may be very high.
3GPP2 C.R1002-A v1.0
177
FTP Traffic Model Packet call delay is measured as the time elapsed between 1
the instant a user arrives in the system and the instant the user departs 2
from the system. 3
WAP Traffic Model Packet call delay is measured as the time elapsed between 4
the instant a WAP request is generated and the instant the next reading time 5
begins. 6
HTTP Traffic Model Packet call delay is measured as the time elapsed between 7
the instant a main page HTTP request is generated and the instant the next 8
reading time begins. 9
Gaming Traffic Model Packet call delay is measured as the time elapsed 10
between the instant a user arrives in the system and the instant the user 11
departs from the system. 12
Due to the fixed simulation time, there may be outstanding not completed packet call at 13
the end of a simulation run. For such a packet call, the end of the packet call is the end 14
of the simulation. 15
Users that do not have even a portion of a packet call during the time statistics is 16
collected (simulation time without warm-up period), are not taken into account for 17
packet call statistics. 18
Average packet call delay per user is computed as the average of all packet call delays 19
of that user during the simulation. 20
Since for each FTP and Gaming user, there is only one packet call, the average 21
packet call delay is the same as the packet call delay. 22
Average packet delay per user is computed as the average of all packet delays during 23
the simulation. 24
FTP Traffic Model Packet delay is defined as the time elapsed between the 25
instant a TCP segment is passed to the transmitter physical layer buffer at 26
the mobile station and the instant all physical layer packets that contain the 27
TCP segment bits are received at a base station. 28
If HARQ operation is implemented, the TCP segment is considered 29
―received‖ when all physical layer packets that contain the TCP 30
segment bits are ACKed or after being transmitted the maximum 31
number of times. 32
WAP Traffic Model Packet delay is defined as the time elapsed between the 33
instant a WAP request is passed to the transmitter physical layer buffer at 34
the mobile station and the instant all physical layer packets that contain the 35
WAP request bits are received at a base station. 36
If HARQ operation is implemented, the WAP request is considered 37
―received‖ when all physical layer packets that contain the WAP 38
request bits are ACKed or after being transmitted the maximum 39
number of times. 40
3GPP2 C.R1002-A v1.0
178
HTTP Traffic Model Packet delay is defined as the time elapsed between the 1
instant an HTTP request (main page request or embedded object request) is 2
passed to the transmitter physical layer buffer at the mobile station and the 3
instant all physical layer packets that contain the HTTP request bits are 4
received at a base station. 5
If HARQ operation is implemented, the HTTP request is considered 6
―received‖ when all physical layer packets that contain the HTTP 7
request bits are ACKed or after being transmitted the maximum 8
number of times. 9
Gaming Traffic Model Packet delay is defined as the time elapsed between the 10
instant a gaming packet is passed to the transmitter physical layer buffer at 11
the mobile station and the instant all physical layer packets that contain the 12
gaming packet bits are received at a base station. 13
If HARQ operation is implemented, a gaming packet is considered 14
―received‖ when all physical layer packets that contain the gaming 15
packet bits are ACKed or after being transmitted the maximum 16
number of times. 17
Average TCP ACK delay per user is calculated as the average of all TCP ACK delays 18
during the simulation. 19
TCP ACK delay is defined as the time elapsed between the instant an TCP ACK is 20
generated at the mobile station and the instant the TCP ACK is received at 21
the base station. 22
If HARQ operation is implemented, the TCP ACK is considered 23
―received‖ when all physical layer packets that contain the TCP ACK 24
bits are ACKed or after being transmitted the maximum number of 25
times. 26
Applicable only to HTTP users. 27
28
3GPP2 C.R1002-A v1.0
179
APPENDIX N: LINK PREDICTION METHODOLOGY FOR UPLINK SYSTEM 1
SIMULATIONS 2
Figure N-1 below outlines the procedure with supporting notation in Table N-1. 3
2
, ,
,
Inputs:
Packet-related information: , , , modulation.
Channel-related information: Channel model, , .
SNR of channel estimator : .
( 1,..., , and 1,..., )
eff i
i p i p
i p
N R E
p P i n
,
Link error prediction algorithm
Output:
use Eqn. 0.1 or 0.2 or 0.3
E eff
,
,
Probability of the packet being in error equals
( ) (Channel model) (Channel model) ,
or, equivalently,
(Channel model), , Channel model
MC E eff P eff D NG
C E eff NG eff
f R
f R
4
Figure N-1: Outline of Equivalent SNR Method. 5
The slots spanned by the transmission include H-ARQ retransmissions, if any, and, 6
therefore, need not be contiguous. The output of the link error prediction algorithm (see 7
Figure N-1) is the effective Eb/N0, ,E eff , for the transmission 1 . This is calculated 8
analytically using Eqn. N-1 for BPSK and N-2 or N-3, for QPSK. 9
1 A block transmission, as referred to here, may consist of one or more H-ARQ retransmissions.
3GPP2 C.R1002-A v1.0
180
Table N-1. Notations used. 1
Variable Description
effR
Effective code rateeff
N
N C
, where effC is the total number of
unique parity bits – excluding systematic bits – transmitted so
far1.
,i pG
Complex channel gain on path p in slot i, which equals ,
,i pj
i pe
,
where ,i p is the magnitude of the channel gain, and ,i p is the
channel phase.
,ˆ
i pG Receiver‘s estimate of the channel gain ,i pG , which will also be
written as ,ˆ
, ,ˆ ˆ i pj
i p i pG e
.
( )P effR Puncturing penalty for a code of effective rate effR .
(Channel model)D Doppler penalty for the given channel model.
(Channel model)NG
Adjustment term, which will be called the Non-Gaussian gain.
This is to account for the fact that the noise introduced due to
demodulation using imperfect channel estimates in non-
Gaussian.
( )MCf x FER for the mother code, when Eb/N0 equals x, on an AWGN
channel.
( , ,Channel model)C efff x R FER for a code of effective rate effR when the short term Eb/N0
equals x on the given channel.
N Number of information bits.
n Total number of slots spanned by the transmission (need not be
contiguous).
S Number of modulation symbols transmitted in each slot.
iE Transmit energy per modulation symbol in slot i.
P Total number of resolvable paths of the channel.
,i p Magnitude of the channel gain on path p in slot i.
1 Note that the effective code rate, by definition, is no smaller than the mother code rate – denoted
MCR – and, in
particular, remains unchanged in the case of Chase combining. In the case where incremental redundancy is used, the formula for effective rate may need some modification to account for the following case: When the
subpacket code rates are very high, and the received SNRs of the subpackets are significantly different, then the
link performance can be worse than that due to a transmission of effective code rate /eff
N N C .
3GPP2 C.R1002-A v1.0
181
Variable Description
2
,i p Interference plus thermal noise power on path p in slot i.
,i p
Equals 2 2
, ,/i p i i pE . Therefore, this is the received modulation
symbol Es/N0 on path p in slot i of the traffic channel before
demodulation. This equals the SNR of the demodulated symbol
only when the channel estimates are perfect.
,i p ―SNR of the channel estimator.‖ See Section 3.1
cN
Number of chips per slot
2
,i p
Variance of the complex additive noise corrupting each pilot
chip. Variance of the real and imaginary parts are each 2
, /2i p
1
Equations (N.1) – (N.3) are all for the case of pilot weighted combining. 2
Effective Eb/N0 for BPSK Modulation: In this case, ,E eff is given by 3
2
2
, ,
1 1
, 4
, , 2
, , ,21 1 , ,
2
2
,
1 1
2
, 2
, , ,
1 1 ,
1
1
n P
i p i p
i p
E effn P
i p i p
i p i p i p
i p i p i p
n P
i i p
i p
n Pi i p
i p i p i p
i p i p
S
N
ES
N E
, (N-1) 4
Effective Eb/N0 for QPSK Modulation with I and Q Branches on Different Walsh 5
Codes: The effective Eb/N0, ,E eff , in this case equals 6
2
2
, ,
1 1
, 4
, , ,2
, ,21 1 , ,
2
2
,
1 1
2
, ,2
, ,
1 1 ,
12
.
12
n P
i p i p
i p
E effn P
i p i p i p
i p i p
i p i p i p
n P
i i p
i p
n Pi i p i p
i p i p
i p i p
S
N
ES
N E
(N-2) 7
Effective Eb/N0 for QPSK Modulation with I and Q Branches on the Same Walsh 8
Code: The effective Eb/N0, ,E eff , in this case equals 9
3GPP2 C.R1002-A v1.0
182
2
2
, ,
1 1
, 4
, , 2
, , ,21 1 , ,
2
2
,
1 1
2
, 2
, , ,
1 1 ,
1
1
n P
i p i p
i p
E effn P
i p i p
i p i p i p
i p i p i p
n P
i i p
i p
n Pi i p
i p i p i p
i p i p
S
N
ES
N E
. (N-3) 1
2
N.1 Definition of Required Terms 3
N.1.1 Channel Estimation SNR, ,i p 4
The receiver‘s estimate ,ˆ
i pG of the channel gain ,i pG on path p in slot i can, in general, be 5
written as 6
, , ,ˆ
i p i p i pG G (N-4) 7
where , ,(0, 1/ )i p i p C . Clearly the variance of ,i p , denoted 1/ ,i p , determines the 8
―quality‖ of the receiver‘s estimate of the channel gain, which leads to the interpretation 9
that ,i p is the SNR of the channel estimator. The value of ,i p depends on the pilot 10
transmission power, the interference plus thermal noise on the pilot channel, and the 11
channel estimator being used. The SNR of specific channel estimators are now provided. 12
For a channel estimator based on one-slot rectangular filter averaging of the pilot symbols 13
2
, ,/P
i p i C i pE N , (N-5) 14
where P
iE is the transmit pilot energy per pilot chip, and CN is the number of chips in the 15
slot and 2
, i p is the variance of the complex additive interference and noise for each pilot 16
chip. 17
In the case of channel estimation based on two-slot rectangular filter averaging, the 18
channel estimates in adjacent slots become correlated. For the purpose of illustration, 19
consider channel estimation based on averaging two successive slots. The channel estimate 20
for slot i is based on averaging the pilot signal from slots (i-1) and i. Suppose transmission 21
takes place in slots i and (i+1). There are two sources of correlation in the channel 22
estimates. Firstly, the channel estimates in the two slots both depend on the channel gain 23
in slot i. Secondly, both channel estimates depend on the additive noise on the pilot in slot 24
i. Two simple approaches are outlined in [28]. One approach captures the correlation due to 25
the channel gains but not the correlation due to the additive noise. 26
For the case where the center tap of the rectangular FIR filter is at the center of slot ( 1)i , 27
so that the channel estimate is based on the pilot signal received in slot i, ( 1)i and ( 1)i , 28
an equivalent expression for the channel estimation SNR is computed as 29
3GPP2 C.R1002-A v1.0
183
, 2 2 22 , 1, 1,
, 1,
1 1
4
1 1 1
4 2
i p
i p i p i p
i p i p p p p
c i i iN E E E
, (N-6) 1
where , 1, ,i p i p i pG G equals the change in channel condition from slot ( 1)i to slot i on 2
path p. The effective Eb/N0 is calculated as before using equation (N-1) or (N-2) or (N-3). In 3
the second approach, the correlation in channel gains and noise are captured and an exact 4
expression for the effective Eb/N0 is calculated (see [28]). Results indicate that the first 5
approach, which is simpler, is accurate and therefore the second approach is not described 6
here. 7
N.1.2 Non-Gaussian Penalty, NG 8
When the channel estimates are very poor – as is the case at very low pilot SNRs –the FER 9
vs effective Eb/N0 curve and the FER curve with perfect demodulation show a small 10
relative shift. This is because the noise introduced by imperfect demodulation is not 11
Gaussian. Therefore, in certain cases, a small adjustment term – called the non-Gaussian 12
gain – may be necessary. Refer to Appendix D in [28] for a more detailed discussion. 13
N.1.3 Reference Curves 14
Once the effective Eb/N0 is computed, the probability of error for the transmission or FER, 15
can be obtained in two alternative ways. In the first method, the probability of error for the 16
transmission or FER is simply , ( ) (Channel model) (Channel model)MC E eff P eff D NGf R 17
i.e., the FER is obtained using the lookup curve MCf after adjusting the effective Eb/N0, 18
,E eff , by applying the Doppler penalty, the puncturing penalty, and the non-Gaussian gain 19
in a manner similar to the methods in Appendix F and [26] Let (Rf x be the FER of a turbo 20
code of effective rate R on an AWGN channel, when Eb/N0 equals x and the modulation is 21
QPSK. The puncturing penalty for the code ( )P R is simply the additional Eb/N0 required 22
for the code to achieve an FER of 0.01, when compared with the mother code. 23
Mathematically, ( )P R is such that 1% 1%( ) ( ) 0.01MC MC R MC Pf E f E R , where 1%
MCE is the 24
Eb/N0 required to achieve an FER of 1% for the mother code on the AWGN channel. The 25
Doppler penalty captures the channel variability during the transmission as compared to an 26
AWGN channel where the channel gain would remain unchanged during the transmission 27
and is obtained in a manner similar to the puncturing penalty. An accurate method for 28
characterizing this penalty due to channel variations would allow the use of a single AWGN 29
reference curve for all cases, while still capturing the effect of noisy pilot using the ESM 30
method. 31
In the second method, short-term FER curves for the given channel model and effective 32
code rate are used to obtain the FER. Here, the puncturing and Doppler penalties are not 33
applied, but the non-Gaussian penalty may still be applicable. 34
The method of generating the short-term curves depends on whether the traffic channel is 35
power controlled or not. The short term FER curve for a given channel model 36
( , , Channel model)C efff x R is obtained from link level simulations, assuming perfect 37
3GPP2 C.R1002-A v1.0
184
channel estimation on each resolvable path of the channel. The goal of this is to 1
characterize the probability of error for a packet transmission over the given channel model 2
when the received Eb/N0 equals x. Of course, since the power of the received signal 3
fluctuates over of the course of the transmission, ( , , Channel model) ( )effC eff Rf x R f x , in 4
general. Since the variation in the received power depends on whether or not the traffic 5
channel is power controlled, the method of generating these short-term curves is different 6
for the two cases. When the traffic channel is not power controlled, the power of the 7
received signal fluctuates in accordance with the corresponding channel model. In this 8
case, the transmit power is kept constant over the course of the transmission, so that 9
( , , Channel model)C efff x R is simply the FER when the Eb/N0 for the packet transmission 10
equals x. By contrast, when the traffic channel is power controlled, the transmit power 11
varies over the course of the transmission, so that the fluctuation in the power of the 12
received signal is a result of the effect of channel variation and the transmit power variation 13
due to power control. Note that since the channel estimation is assumed to be perfect, the 14
degradation due to imperfect channel estimation is not captured in these short-term 15
curves. 16
With combining of multiple Hybrid-ARQ transmissions, the effect of diversity due to the 17
multiple transmissions would need to be captured in the reference curves1. The most 18
precise method would be to have reference curves parameterized based on channel model, 19
code rate, receive antenna diversity order and number of transmissions. However, to keep 20
the number of reference curves reasonable, some simplifications are possible. 21
22
a) At low speeds and two receive antennas, a single short-term curve may suffice. For 23
example, for Channel Model A (Pedestrian A) with two receive antennas, a single 24
reference curve per code rate is sufficient and multiple reference curves based on 25
number of transmissions is not needed. 26
b) At medium speeds, if the overall ―diversity order‖ is high, (e.g., large number of 27
multipaths in the channel models and/or the use of receive diversity), then again a 28
single curve may be sufficiently accurate in predicting performance. For example, for 29
Channel Model B (Pedestrian B) and C (Vehicular A) with two receive antennas, a single 30
curve per code rate is sufficient and multiple reference curves based on number of 31
transmissions is not needed. 32
c) Furthermore, if the diversity order is low and the channel variation is high, then 33
some simplifications can still be used to minimize the number of reference curves. For 34
example, the reference curve for the first transmission with two receive antennas is the 35
same as that for the second transmission with only one receive antenna. 36
37
1 Note that the diversity obtained from multiple transmissions reduces the extent of channel
variations.
3GPP2 C.R1002-A v1.0
185
APPENDIX O: REVERSE LINK HYBRID ARQ: LINK ERROR PREDICTION 1
METHODOLOGY BASED ON CONVEX METRIC 2
A link error prediction (LEP) method that can predict, by using single AWGN reference 3
curve, the performance of different fading channels, the performance of multiple re-4
transmissions with power imbalances and the performance of different frame durations is 5
described in this text. The method uses the Equivalent SNR Method (ESM) [3] to calculate 6
the effects of a weak pilot signal. The main idea of this method is using a new SNR metric 7
based on Shannon‘s channel capacity formula to reflect the performance loss due to 8
channel gain variations. The new metric can predict the performance at different fading 9
channels, and the diversity gain due to multiple transmissions. 10
O.1 Convex Metric based on Channel Capacity Formula 11
The quasi-static method (QSM) [1] obtains the average frame tb NE / by averaging tb NE / 12
per slot and by looking up the corresponding frame FER from a reference curve. However, 13
having the same average frame tb NE / , the channel with less variation shows better 14
performance than the channel with more variation. Consequently, AWGN channels and 15
slowly varying channels have better error performance compared to fast varying channels. 16
Since the existing QSM does not take this second order statistics into account, a new 17
performance metric, which is based on Shannon‘s channel capacity formula, is proposed. 18
Consider a deterministic fading channel where we know the channel response. Piece-wise 19
AWGN channels can then approximate the fading channel. Then the channel capacity of 20
the frame can be calculated as 21
M
m
mCM
C1
1 (O.1-1) 22
where C is the channel capacity, M is the number of segments that have approximately 23
constant channel response, and Cm is the channel capacity of m-th segment. If we use 24
Gaussian signaling, Cm is expressed by 25
m
m
mN
SC 1log (O.1-2) 26
where Sm and Nm are the channel energy and AWGN noise variance of m-th segment, 27
respectively. By combining equation (O.1-1) and (O.1-2), the channel capacity is expressed 28
by 29
M
m m
m
M
m m
m
N
S
M
N
S
MC
1
1
1log1
1log1
(O.1-3) 30
As we can see from equation (O.1-3), C is maximized if all mm NS / have the same value. 31
If the variance of mm NS / increases, the value of C decreases. This formula can explain 32
3GPP2 C.R1002-A v1.0
186
the performance degradation due to increased mobile speed, and diversity gain. Diversity 1
gain is obtained by making the channel variations smaller after combining multiple frames. 2
The effective SNR, ts NE / , of a frame has one to one function relationships with channel 3
capacity C, and it is calculated from equation (O.1-2) as 4
121log 2
C
t
s
t
s
N
E
N
EC (O.1-4) 5
In reality, the capacity formula in equation (O.1-4) may not be directly applied to different 6
EP sizes and channel code rates. The size of a segment in our simulation is one PCG (1.25 7
ms). In case of high speed mobile, channel gain within a segment is not constant and 8
channel variations within a segment are not properly penalized. This can be avoided by 9
making the segment size small enough so that the channel gain is constant over a segment. 10
Furthermore, non-ideal channel interleaver causes additional performance degradation at 11
channels with high gain variations. After all, practical systems do not exactly follow the 12
capacity formula, and we therefore need an adjustment factor in the equation. The modified 13
metric is expressed as 14
QN
E
N
EQC
C
t
s
t
s 121log2
(O.1-5) 15
where Q is a positive real number. A higher value of Q means more convexity of the 16
capacity formula and more penalties for channel gain variations. Thus Q can be considered 17
as a penalty factor. 18
If we use binary signaling, the mutual information formula in equation (O.1-2) needs to be 19
changed. Let 2/oN be the noise variance of an AWGN channel. Let E be the transmitted 20
symbol energy, i.e., if s is SNR in dBs then 10/10/ s
ts NE . Assume we transmit a 21
channel bit, X, with signal to noise ratio s dB through the AWGN channel and on the other 22
end of the channel we receive a Gaussian random variable, Y. Let Z be the LLR (log 23
likelihood ratio) value of Y, that is 24
EXY
EXYz
|Pr
|Prln . (O.1-6) 25
Then the variance of Z is 10/2 108 s and the mean of Z equals 2/2 . Using this and 26
the definition of the mutual information, we obtain, after some computations, the following 27
expression 28
.1log2
11)();( 2
2/)2/( 222
dveeJYXIvv
. ( O.1-7) 29
Thus there is a one- to- one correspondence between the mutual information and the signal 30
to noise ratio. There is an easy way to compute the function )(xJ by using the following 31
approximation 32
x 1.6363 ifxexp(a-1
1.6363x if3
2 )
,)(
22
2
2
1
2
1
3
1
dxcxb
xcxbxaxJ ( O.1-8) 33
3GPP2 C.R1002-A v1.0
187
where 209252.0,0421061.0 11 ba and 00640081.01 c for the first approximation, and where 1
0822054.0,142675.0,00181491.0 222 cba and 0549608.02 d for the second approximation. 2
In a similar fashion, one can determine the inverse of )(xJ by using the following 3
approximation 4
1
,)(
444
33
2
31
y 0.3646 ify1))-(yln(
0.3646 y0 if
cba
ycybyayJ ( O.1-9) 5
where 214217.0,09542.1 33 ba and 33727.21 c for the first approximation, and where 6
386013.0,706692.0 44 ba and 75017.14 c for the second approximation. 7
We can express the above equation as 8
t
s
BN
EfC ( O.1-10) 9
where CB means binary capacity. The function f( ) is a one-to-one function. Like equation 10
(O.1-5), we have an adjustment factor Q. The modified formula is 11
Q
CfN
E
N
EQfC B
t
s
t
sB
11
( O.1-11) 12
The channel capacity of QPSK modulation is derived in a similar fashion and is 13
approximated by 14
t
s
QN
EQfC
22 ( O.1-12) 15
Approximate channel capacities for BPSK and QPSK signalling are plotted in Figure O.1. 16
0 1 2 3 4 50
0.5
1
1.5
2
2.5
3Channel capacity: gaussian signaling, QPSK, BPSK
Es/No (linear)
Bits/H
z
BPSK Gaussian signalingQPSK
17
Figure O-1. Approximate channel capacities at BPSK and QPSK signaling (Gaussian 18
signaling case is plotted as a reference). 19
3GPP2 C.R1002-A v1.0
188
The key point of using equation (O.1-5) or (O.1-11) is that we have metric that can account 1
the diversity gain or the performance loss due to channel gain variations. And we may use 2
one AWGN reference curve for all fading channels and multiple re-transmissions of H-ARQ 3
packets. However, it does not cover the effect of weak pilot at reverse link. There was 4
another contribution (Equivalent SNR Method: ESM) from Lucent Technologies that 5
provides the formula to calculate the effective ts NE after distortions due to weak pilot 6
[3]. By using ESM, we may not need multiple reference curves for different pilot SNR‘s. As a 7
result, only one reference AWGN reference curve is necessary for wide ranges of pilot SNR, 8
channel variations and multiple re-transmissions. The enhanced methodology that 9
combines ESM and convex metric is named as ‗‖ESM based on Convex Metric” (ECM), and is 10
described in the next section. 11
O.2 Equivalent SNR Method based on Convex Metric (ECM) 12
This scheme combines ESM [3] with the convex metric based on capacity formula. The 13
detailed procedures are explained in the following. 14
O.2.1 Overview of the Procedure 15
Assume that M is the number of slots, S is number of modulation symbols in each slot, and 16
N is number of information bits. We have all other variables (like channel response, 17
transmission power, noise variances per slot, etc.) available. The ESM method combines all 18
these information to produce effective tb NE per frame and compare it to reference curve 19
(that is obtained on specific fading channel with perfect pilot information). 20
ECM uses the same formula as ESM, but calculates the effective ts NE per slot. Thus we 21
have the following M effective ts NE values per slot. 22
Mt
s
t
s
t
s
t
s
N
E
N
E
N
E
N
E
,,,,
321
(O.2.1-1) 23
Then the channel capacity of each slot Cm is (assume we use the formula in equation (O.1-24
5). If we use BPSK or QPSK modulation, we use the formulae in equation ( O.1-11) or ( O.1-25
12), respectively) 26
mt
s
mN
EQC 1log 2
(O.2.1-2) 27
Then we average Cm so that the final capacity C is (from equation (O.1-1)). 28
Ft
sM
m
mN
EQC
MC 1log
12
1
(O-3) 29
Then the new effective Fts NE can be obtained as 30
QNt
EsC
F
12
(O.2.1-4) 31
3GPP2 C.R1002-A v1.0
189
Let Ftb NE / , is final effective Eb/Nt, then
Ftb NE / can be obtained as 1
Ft
s
s
Ft
b
N
EN
N
E
(O.2.1-5) 2
where sN is the number of modulation symbols per information bit. Then we can obtain 3
the FER by applying Ftb NE to an AWGN reference curve. This procedure is illustrated 4
in Figure O.2. 5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Figure O-2. Overview of ECM procedure (in step 2, we assume Gaussian signaling. If 21
BPSK or QPSK are used, channel capacity formula in equation (O.1-11) or (O.1-12) has to 22
be used). 23
O.2.2 Detailed Procedure 24
The procedure is described with a specific example for better understanding. For 25
convenience, all the following notations are defined in Table N-1 in [3]. 26
Let us consider 153.6 kbps R-SCH on channel model A (3 km/h and single path fading). 27
There are two receiver antennas (P=2), and 16 slots per frame (n=16). The number of 28
information bits in an encoder packet is 3072 per frame (N=3072), and the number of 29
modulation symbols per slot is 768 (S=768). Let us assume that we have a fixed traffic-to-30
pilot ratio (TPR), and let the energy of the transmitted pilot energy per modulation symbol 31
be 1. Then the transmitted energy per modulation symbol in slot i, iE , is TPR. Actually, the 32
transmitted power is not constant over the frame because of power control. However, we 33
assume that power control is part of the fading channel gain variations, and ,i p includes 34
Step 1. Calculate the equivalent (Es/Nt)m per m-th slot that has weak pilot effects.
(Es/Nt)1
(Es/Nt)2
(Es/Nt)3
(Es/Nt)4
Step 2. Calculate the channel capacity C of the frame.
mt
s
m
m
mN
EQCCC 1log
4
1 4
1
where ,
Step 3. Calculate the effective (Es/Nt) of the frame, and obtain (Eb/Nt)
QN
E C
t
s 12
Step 4. Obtain the FER by applying (Eb/Nt) to an AWGN reference curve
3GPP2 C.R1002-A v1.0
190
the power control effects. We can have values of ,i p in the system level simulation. The 1
noise variance 2
,i p includes the thermal noise and multi-path interference. The variance of 2
thermal noise can be calculated from the pilot SNR set point. In this example, there are two 3
chips per modulation symbol. Thus, the pilot energy per chip is 0.5. If the pilot set point 4
per antenna is –20 dB, then the thermal noise variance per chip per antenna is 50. When 5
we have more than one path per antenna, we add the effect of multi-path interference to 6
2
,i p . Even though the multi-path interference does not have Gaussian statistics, we model 7
it as Gaussian random variables. The SNR of the channel estimator ,i p is derived in 8
section N.1.1 [3]. 9
There is another simple modeling of pilot estimation SNR, and it is described in the 10
following. Given Doppler frequency fd, the coherent accumulation loss (fd) is expressed as 11
2
)sin()(
Tf
Tff
d
dd
(O.2.2-1) 12
Where T is the integration time duration. T is 0.0025 in our simulation (2 PCG). Doppler 13
frequency fd is 14
cos
C
C
dv
fvf (O.2.2-2) 15
Where v is mobile speed, vC is speed of light, fC is carrier frequency, is angle of arrival. If 16
we assume has uniform distribution, the mean loss E[] can be calculated per mobile 17
speed. The calculated value of E[] is 0.623 for 120 Km/h mobile, and 0.969 for 30 Km/h 18
mobile. Thus the channel estimation SNR is 30720.623(pilot SNR per chip) = 1914(pilot 19
SNR per chip) for 120 Km/h, and 2977(pilot SNR per chip) for 30 Km/h mobiles. For 20
channel model A and B, we may not need to put penalty on. ,i p . 21
Then we apply the equation (N.1 in [3]) to calculate effective Es/Nt per slot. In this case the 22
effective ts NE of i-th slot can be calculated as 23
24
P
p
pipipi
pi
pii
P
p
pii
it
s
E
E
N
E
1
,,
2
,
,
2
,
2
1
2
,
1
(O.2.2-3) 25
For the convenience of notation, we define the signal component iA and noise component 26
iN on i-th slot as 27
3GPP2 C.R1002-A v1.0
191
P
p
pipipi
pi
pii
i
P
p
piii
EN
EA
1
,,
2
,
,
2
,
1
2
,
1
(O.2.2-4) 1
Then effective ts NE per i-th slot is calculated as 2
i
i
it
s
N
A
N
E2
(O.2.2-5) 3
Then we follow the procedure described in section 3.1. 4
5
O.2.3 Combining Procedure for H-ARQ 6
Now, we discuss the combining procedure for H-ARQ (Chase combining and IR). First we 7
consider Chase combining case. Lets define kiA ,and
kiN , as the signal component and 8
noise component of k-th transmission on i-th slot. Then, iA and iN after Chase 9
combining are calculated as 10
K
k
kii
K
k
kii NNAA1
,
1
, , (O.2.3-1) 11
where K is total number of transmissions. Then we apply equation (O.2.2-5) for calculating 12
effective ts NE of the i-th slot. 13
Consider the case of IR.1 Suppose there are total K transmissions and each transmission 14
has h(k) number of slots (where k = 1,2,…,K), and there is no overlapping of transmitted 15
symbols. Then, we apply ECM method as if there is a single transmission of KL slots where 16
K
k
K khL1
)( . ( O.2.3-2) 17
The reference channel code rate is the effective code rate after all K transmissions. If there 18
are overlapping symbols, we apply the Chase combining metric for that overlapping portion. 19
Simulations and the range of Q values are described in the next section. 20
21
1 Simulation results based on two transmission IR shows that we can have same Q values if the
received power set-point difference of both transmissions is less than 3 dB. If the power difference is
larger than 3 dB, adjustments for Q values are required.
3GPP2 C.R1002-A v1.0
192
APPENDIX P: PILOT SINR ESTIMATION FOR POWER-CONTROL COMMAND UPDATE IN 1
LINK-LEVEL SIMULATIONS 2
In link-level simulations, the following method can be applied to estimate pilot Ec/Nt that 3
is used to update the power-control command. The corresponding procedure for system-4
level power-control update is specified in section 3.2.3. 5
The pilot decovered symbols may be modeled as 6
ilp
j
ilpcpilpilpcpilP neENnANy il
,,,,,,,,,,,,,
(P-1) 7
where NP is the length of the pilot cover code; Ac,p,l,i, Ec,p,l,i and l,i denote the complex chip 8
amplitude, energy and phase associated with the i-th decovered pilot symbol for the l-th 9
RAKE finger output; np,l,i denotes the corresponding interference which can be modeled as a 10
Gaussian random variable with zero mean and a variance of NpNt,l, where Nt,l denotes the 11
interference variance per chip for the l-th RAKE finger output. Assuming that channel 12
conditions remain approximately constant over Ns = K Np chips (i.e. the Ec/Nt averaging 13
length specified in appendix J), the estimates of the complex amplitudelpcA ,,
ˆ , noise 14
variance ltN ,
ˆ over the interval of symbols [n, n+ K) can be obtained as follows: 15
1
,,,,
^ 1 Kn
ni
ilp
s
lpc yN
A ,
1 2
,,
^
,,,
^ 1 Kn
ni
lpcpilp
s
lt ANyN
N . (P-2) 16
From this, the per-path, per-chip pilot SINR estimate
estt
pc
N
E
,used for power control may 17
be obtained as 18
L
l lt
lpc
estt
pc
N
A
N
E
1 ,
2
,,,
ˆ
ˆ
(P-3) 19
This expression is consistent with specifications in section 3.2.3. 20
21
3GPP2 C.R1002-A v1.0
193
APPENDIX Q: 1XEV-DV REVERSE LINK SIMULATION AND SCHEDULER PROCEDURES 1
Q.1.1 Mobile station Requirements and Procedures 2
IS-2000 Release C-capable mobile stations (MS) should at least support the simultaneous 3
operation of the following channels: 4
1. Reverse Fundamental Channel (R-FCH): When a voice-only MS has an active voice-call, 5
it is carried on the R-FCH. For data-only MS, R-FCH carries signaling and data. R-FCH 6
channel frame size, coding, modulation and interleaving are specified in [29]. For 7
calibration purposes, we restrict the R-FCH use only Radio Configuration (RC) 3 and 8
the 20-ms frame format. 9
2. Reverse Supplemental Channel (R-SCH): The MS supports one R-SCH for packet data 10
transmissions. The R-SCH uses rates specified by RC3 in [29]. For the purpose of 11
calibration, we restrict the use of 9.6-kbps, 20-ms convolutional coded frame format 12
and (19.2, 38.4, 76.8, 153.6, 307.2) kbps turbo-coded, 20-ms frame format. 13
The following procedures are followed by IS-2000 Release C MSs: 14
1. Multiple Channel Adjustment Gain: When the R-FCH and the R-SCH are 15
simultaneously active, multiple channel gain table adjustment is performed to maintain 16
correct transmission power of the R-FCH. The traffic to pilot (T/P) ratios for all channel 17
rate are also specified in the table 2.1.2.2.3.4-1 as Nominal Attribute Gain values [29]. 18
The Pilot Reference Levels used for multi-channel gain adjustment are specified in 19
Section 3 [29]. 20
2. Discontinuous Transmission and Variable Supplemental Adjustment Gain: The MS may 21
be assigned an R-SCH rate by a scheduler during each scheduling period. When the MS 22
is not assigned an R-SCH rate, it will not transmit anything on the R-SCH. If the MS is 23
assigned to transmit on the R-SCH, but it does not have any data or sufficient power to 24
transmit at the assigned rate, it disables transmission (DTX) on the R-SCH. If the 25
system allows it, the MS may be transmitting on the R-SCH at a rate lower than the 26
assigned one autonomously. This variable-rate R-SCH operation has to be accompanied 27
by the variable rate SCH gain adjustment as specified in [29]. R-FCH T/P is adjusted 28
assuming the received pilot SNR is high enough to support the assigned rate on R-SCH. 29
3. Overhead transmission of R-CQICH: A data-only MS transmits extra power 30
continuously at a fixed CQICH-to-pilot (C/P) ratio with multi-channel gain adjustment 31
performed to maintain correct transmission power of the R-CQICH. (C/P) value is 32
different for MS in soft-handoff from those not in soft handoff and is given in Section 3. 33
4. Closed-loop Power Control (PC) command: MS receives one PC command per PCG (at a 34
rate of 800Hz) from all base stations (BSs) in the MS‘s Active Set. Pilot power is updated 35
by +-1 dB based on ―Or-of-Downs‖ rule, after respectively combining of the PC 36
commands from co-located BSs (sectors in a given cell). 37
5. Rate request is done with one of two methods. Both methods are to be simulated. 38
Method a (5 ms): Supplemental Channel Request Mini Message (SCRMM) on a 39
5-ms R-FCH: Signaling overhead due to the SCRMM should be modeled. This is 40
3GPP2 C.R1002-A v1.0
194
the only signaling overhead modeled on RL on R-FCH. Each SCRMM 1
transmission is 24 bits (or 48 bits with the physical layer frame overhead in 2
each 5-ms FCH frame at 9.6 kbps). 3
The MS can send the SCRMM in any periodic interval of 5 ms. If a 5-ms SCRMM needs 4
to be transmitted, the MS interrupts its transmission of the current 20-ms R-FCH 5
frame, and instead sends a 5-ms frame on the R-FCH. After the 5-ms frame is sent, any 6
remaining time in the 20-ms period on the R-FCH is not transmitted. The discontinued 7
transmission of the 20-ms R-FCH is re-established at the start of next 20-ms frame. 8
The frame error on the 5-ms R-FCH frames should be simulated. If a 5-ms R-FCH frame 9
is in error, the SCRMM carried on this frame is considered lost. No LAC layer 10
retransmission is simulated. 11
Method b (20 ms): Supplemental Channel Request Message (SCRM) on a 20-ms 12
R-FCH: Signaling overhead due to the SCRM should be modeled. This is the only 13
signaling overhead message on the R-FCH that is modeled on the RL. One 20-14
ms 9.6-kbps R-FCH frame transmission could be used to model the SCRM 15
overhead1. No data is transmitted on the R-FCH when the SCRM is sent. 16
The frame error on the 20-ms R-FCH frames should be simulated. If a 20-ms R-FCH 17
frame is in error, the SCRM carried on this frame is considered lost. No LAC layer 18
retransmission is simulated.‖ 19
The following information shall be reported by the MS to the BS on each SCRM/SCRMM 20
transmission: 21
Maximum Requested Rate: It is the maximum data rate a MS is capable of 22
transmitting at the current channel conditions leaving headroom for fast 23
channel variations: 24
25
max 9.6
: ( )* ( )*
(9.6 )( ) arg max (1 ( / ) (( / ) / ) )
( )
( x) /
i
R kR
R Pref R NormAvPiTx PCG
Pref kR power T P T P C P
Pref R
Tx ma Headroom_Req
26
(Q.1.1-1) 27
1
( )( ) (1 ) ( )
( )
ii Headroom Headroom i
TxPiPwr PCGNormAvPiTx PCG NormAvPiTx PCG
Pref Rassigned 28
, (Q.1.1-2) 29
where Pref(R) is the ―Pilot Reference Level‖ value specified in the Attribute Gain 30
Table in [29], )( iPCGTxPiPwr is the actual transmit pilot power after power 31
1 With LAC layer overhead included, minimum SCRM payload is 144 bits. This can be transmitted
over one 20-ms 9.6-kbps channel.
3GPP2 C.R1002-A v1.0
195
constraints on the MS side are applied in case of power outage. 1
)( iPCGNormAvPiTx is the normalized average transmit pilot power. 2
Queue Information: For simulation purposes, we can assume that perfect queue 3
information (at the time of transmission of SCRMM) shall be reported. 4
6. MS receives grant information by one of the two following methods: 5
Method a: ESCAMM from BS on 5-ms F-DCCH with rate assignment for 6
specified scheduling duration. Method a is used for grant whenever method a is 7
used for requesting. 8
Method b: ESCAM from BS on F-PDCH with rate assignment for specified 9
scheduling duration. Method b for grant is used whenever method b is used for 10
requesting. 11
A 1% frame error rate is assumed for the 5-ms F-DCCH frames. The ESCAM carried 12
on F-PDCH is considered error free due to HARQ retransmission on F-PDCH. 13
The assignment delays are different depending on which method is used for rate 14
grant. 15
During the scheduled duration, the following procedures are performed: 16
At the beginning of each 20-ms R-FCH frame, the MS transmits data at the 17
rate of 9600 bps if it has some data in its buffer. Otherwise, the MS sends a 18
null R-FCH frame at a rate of 1500 bps. 19
The MS transmits at the assigned R-SCH rate in a given 20-ms period if the 20
MS has more data than can be carried on the R-FCH and if the MS has 21
decided that it would have sufficient power to transmit at the assigned rate 22
(keeping headroom for channel variations). Otherwise, there is no 23
transmission on the R-SCH during the frame. The MS decides that it has 24
sufficient power to transmit on the R-SCH at the assigned rate R in a given 25
20-ms period Encode_Delay before the beginning of that 20-ms period if the 26
following equation is satisfied: 27
( ) ( )( )* ( ) 1 ( / ) (( / ) ( / ))
( ) _FCH
FCH
i R R
Pref R Tx maxPref R NormAvPiTx PCG T P T P C P
Pref R Headroom Tx
28
(Q.1.1-3) 29
Even if the MS interrupts the 20 R-FCH frame transmission to transmit a 5 30
ms SCRMM transmission on R-FCH, it uses T/P ratio corresponding to a 31
9.6kbps transmission on a 20 ms frame (3.75 dB) to determine whether it 32
should DTX on the R-SCH. The DTX determination is done once every frame, 33
Encode_Delay PCGs before the R-SCH transmission. If the MS disables 34
transmission on the R-SCH, it transmits at the following power: 35
36
3GPP2 C.R1002-A v1.0
196
( )( ) ( ) 1 (( / ) ( / ))
( )FCH
FCHi i R
Pref RTxPwr PCG PiTxPwr PCG T P C P
Pref R
(Q.1.1-4) 1
A MS encodes the transmission frame Encode_Delay before the actual2
transmission. 3
If there is less available data than what can be transmitted at the assigned 4
rate, the frame is padded with 0‘s and transmitted at the assigned rate. Only 5
the actual data is counted towards system throughput. 6
7
7. Regardless of the grant, a data-only MS transmits at the rate of 9.6 kbps on R-FCH if 8
There is some data in the MS‘s buffer (as determined in Step 5 above), or, 9
There is SCRM transmission on R-FCH (see Step 4 above). 10
When there is no data in the MS‘s buffer and when no SCRM transmission is due, 11
the MS transmits a null 20-ms frame at 1.5 kbps on the R-FCH for effective outer-12
loop PC. 13
Q.1.2 Base Station Requirements and Procedures 14
The BS is required to perform the following essential functions: 15
1. Decoding of R-FCH/R-SCH: When there are multiple traffic channels transmitted by the 16
MS simultaneously, Eb/Nt from other channels should be removed from the total 17
received Eb/Nt before the FER table-lookup is performed. Let the total propagation loss 18
at ith PCG be denoted by PL. The total received power is given by: 19
20
( ) Pr (( ) * ( ) 1 ( / ) (( / )
( ) Pr ( )
)( / ))
FCH
tx FCH
i i Rtx R
Pref R ef RRxPwr PCG PL PiTxPwr PCG T P T P
Pref R ef RC P
21
22
(Q.1.2-1) 23
where R is the assigned R-SCH rate and Rtx = {0, R} is the transmitted rate on R-SCH. In 24
order to reduce the simulation complexity and the number of link-level curves, the 25
following decoding method shall be used: 26
Decoding of R-FCH: Derive the R-FCH and consider the following received Eb/Nt for 27
R-FCH decoding: 28
( )( ) * ( ) 1 ( / )
( ) FCH
FCHi i R
Pref RRxFCHPwr PCG PL PiTxPwr PCG T P
Pref R
(Q.1.2-2) 29
Note that the FCH Eb/Nt considered is the same as received. Eb/Nt computed using 30
the above RxFCHPwr(PCGi) can be used directly to read the erasure probability from 31
the link-level curve corresponding to FCHRPT )/( . Also note that the above 32
3GPP2 C.R1002-A v1.0
197
methodology is conservative since better pilot SNR will slightly improve FCH 1
decoding performance. 2
Decoding of R-SCH: Since the MS either transmits at the assigned rate or DTX its 3
transmission, the following received Eb/Nt for R-SCH decoding is used when MS 4
transmits on that channel: 5
txRii PTPCGPiTxPwrPLPCGRxSCHPwr )/(1)(*)( (Q.1.2-3) 6
Eb/Nt computed using the above RxSCHPwr(PCGi) can be used directly to read 7
erasure probability from the link-level curve corresponding to txRPT )/( . 8
2. Power-control: Power control in a CDMA system is essential to maintain the desired 9
quality of service (QoS). In IS-2000, the RL pilot channel (R-PICH) of each MS is closed-10
loop power controlled to a desired threshold. At the BS, this threshold, called power 11
control set point, is compared against the received Ecp/Nt to generate power control 12
command (closed-loop PC), where Ecp is the pilot channel energy per chip. To achieve 13
the desired QoS on the traffic channel, the threshold at the BS is changed with erasures 14
on the traffic channel, and has to be adjusted when the data rate changes. Set point 15
corrections occurs due to: 16
Outer-loop power control: The power control set point is corrected based on 17
erasures of the R-FCH. A data-only MS while transmitting data and signaling, sends 18
a frame at 9.6 kbps or an null frame at 1.5 kbps on the R-FCH. Outer loop power 19
control does not depend on the R-SCH erasure performance. 20
A delay of 20 ms is used before the power control set-point correction is applied. So 21
for a frame decode at time t (corresponding to a frame boundary of 20 ms), the 22
corresponding set-point correction is applied at t+20. In case of a 5 ms SCRMM 23
transmission interrupting 20 ms R-FCH transmission, the power control set-point 24
correction is still applied at 20 ms frame decode boundary. A 5 ms R-FCH erasure 25
detection affects the outer-loop correction in the same way as a 20 ms R-FCH 26
erasure detection. 27
Rate Transitions: Different data rate on the R-SCH requires different optimal set 28
point of the reverse pilot channel. When data rate changes on the R-SCH, the BS 29
changes the MS‘s received Ecp/Nt by the Pilot Reference Levels (Pref(R)) difference 30
between the current and the next R-SCH data rate. The Pilot Reference Level for a 31
given data rate R, Pref(R), is specified in the Gain Table in C.S0002-C. Since the 32
closed-loop power control brings the received pilot Ecp/Nt to the set point, the BS 33
adjusts the outer loop set point according to the next assigned R-SCH data rate: 34
( ) ( )Pref Rnew Pref Rold (Q.1.2-4) 35
Set point adjustment is done PCGs in advance of the new R-SCH data rate if 36
Rnew > Rold. Otherwise, this adjustment occurs at the R-SCH frame boundary. The 37
pilot power thus ramps up or down to the correct level approximately in 1 dB step 38
sizes of the closed loop (Figure Q-1). 39
40
3GPP2 C.R1002-A v1.0
198
1
Pref
(R1) - P
ref(R
0) P
ref(R
2) - P
ref(R
1)
BTS Rx
Pilot Power
MS Rate
Setpoint
(BSC)
R1
> R0
R0
R2
< R1
2
Figure Q-1: Set point adjustment due to rate transitions on R-SCH 3
4
Q.1.3 Scheduler Requirements and Procedures 5
The following assumptions are used for the scheduler and various parameters associated 6
with scheduling: 7
1. Centralized Scheduling: The scheduler is co-located with the BSC, and is responsible 8
for simultaneous scheduling of MSs across multiple cells. For system simulation with 9
19 cells, one scheduler for simultaneous scheduling of all 19 cells shall be used. 10
2. Synchronous Scheduling: All R-SCH data rate transmissions are time aligned. All data 11
rate assignments are for the duration of one scheduling period, which is time aligned 12
for all the MSs in the system. The scheduling duration is denoted SCH_PRD. The 13
scheduling timeline is discussed in 0. 14
3. Voice and Autonomous R-SCH transmissions: Before allocating capacity to 15
transmissions on R-SCH through rate assignments, the scheduler looks at the pending 16
rate requests from the MSs and discounts for voice and autonomous transmissions in a 17
given cell. Details of the scheduler are in 0. 18
4. Rate Request Delay: The uplink request delay associated with rate requesting via 19
SCRM/SCRMM is denoted as D_RL(request). It is the delay from the time the request is 20
sent to when it is available to the scheduler. D_RL(request) includes delay segments for 21
over-the-air transmission of the request, decode time of the request at the cells, and 22
backhaul delay from the cells to the BSC, and is modeled as a uniformly distributed 23
random variable (see Appendix R). 24
5. Rate Assignment Delay: The downlink assignment delay associated with rate 25
assignment via ESCAM/ESCAMM is denoted as D_FL(assign). It is the time between the 26
moment the rate decision is made and the time the MS receiving the resultant 27
3GPP2 C.R1002-A v1.0
199
assignment. D_FL(assign) includes backhaul delay from the scheduler to the cells, over-1
the-air transmission time of the assignment (based on method chosen), and its decode 2
time at the MS and is modeled as a random variable as described in Appendix R. 3
6. Available Ecp/Nt Measurement: The Ecp/Nt measurement used in the scheduler shall 4
be that available at the last frame boundary before action time minus 31 PCG‘s and 68 5
PCG‘s for Method a and Method b, respectively. See Figure Q-2 below. 6
7
8
Figure Q-2 Scheduling Delay Timing 9
Q.1.3.1 Scheduling, Rate Assignment and Transmission Timeline 10
Given the assumed synchronous scheduling, most events related to request, grant and 11
transmission are periodic with period SCH_PRD. 12
Figure Q-3 illustrates the timing diagram of a rate request, scheduling and a rate 13
allocation. 14
3GPP2 C.R1002-A v1.0
200
Rate
Request R
assig
n(n
)
Trasmission
Turnaround Time
Scheduling
Period
BSC
(Sched
uler)
MobileSCRMM
D_RL(request)
Scheduling
Rtx=Rassign(n)
D_FL(assign)
ESCAMM
Rassig
n(n
+1)
Rtx=Rassign(n+1)
1
Figure Q-3: Parameters associated in mobile station scheduling on RL 2
The following characterizes the simulation timeline: 3
Scheduling Timing: The scheduler operates once every scheduling period. If the first 4
scheduling decision is performed at it , then scheduler operates at 5
..._2,_, PRDSCHtPRDSCHtt iii 6
Scheduled Rate Transmissions: Given that the MSs have to be notified of the 7
scheduling decisions with sufficient lead-time, a scheduling decision has to be 8
reached at Action Time of the ESCAM/ESCAMM message minus a fixed delay, 9
ActionTimeDelay. The values of ActionTimeDelay for Methods a and b are given in 10
Section 0 to ensure that most MSs receive the ESCAM/ESCAMM messages with 11
high probability. 12
If the ESCAM/ESCAMM message is received after (Action Time – 2 PCG), it is 13
considered lost. 14
MS R-SCH Rate Requests: R-SCH rate requests are triggered as described below: 15
Before the beginning of each SCRM/SCRMM frame encode boundary, the MS checks if 16
either of the following three conditions are satisfied: 17
1. New data arrives and data in the MS‘s buffer exceeds a certain buffer depth 18
(BUF_DEPTH), and the MS has sufficient power to transmit at a non-zero rate; OR 19
3GPP2 C.R1002-A v1.0
201
2. If the last SCRM/SCRMM was sent at time i , and the current time is greater than 1
or equal to PRDSCHi _ , and if the MS has data in its buffer that exceeds the 2
BUF_DEPTH, and the MS has sufficient power to transmit at a non-zero rate; OR 3
3. If the last SCRM/SCRMM was sent at time i , and the current time is greater than 4
or equal to PRDSCHi _ , and if the current assigned rate at the MS side based 5
on received ESCAMM/ESCAM is non-zero (irrespective of the fact that the MS may 6
not have data or power to request a non-zero rate). ―Current assigned rate‖ is the 7
assigned rate applicable for the current rate transmission. If no ESCAM is received 8
for the current scheduled duration, then the assigned rate is considered 0. The rate 9
assigned in the ESCAM/ESCAMM message with Action Time at some later time 10
takes effect after the Action Time. 11
If either of the above three conditions are satisfied, the MS sends a SCRMM/SCRM rate 12
request. A SCRM/SCRMM request made at i is made available to the scheduler after a 13
random delay at )(_ requestRLDi . To initialize the first SCRMM transmissions of 14
the various MSs, last SCRMM/SCRM sent time i is selected from an uniformly 15
generated random variable that uniformly picks one frame in the first scheduling 16
period. 17
Q.1.3.2 Scheduler Description and Procedures 18
There is one centralized scheduler for the entire system under consideration (19 cells or 57 19
sectors). The scheduler maintains a list of all MSs in the system and BSs in each MS‘s 20
Active Set. Associated with each MS, the scheduler stores estimate of MS‘s queue size ( Q ) 21
and maximum scheduled rate (Rmax(s)). 22
The queue size estimate Q is updated after any of the following events happen: 23
1. A SCRMM/SCRM is received: SCRMM/SCRM is received after a delay of 24
D_RL(request). Q is updated to: 25
SCRMMinreportedSizeQueueQ ˆ (Q.1.3.2-1) 26
If the SCRMM/SCRM is lost, the scheduler uses the previous (and the latest) 27
information it has. 28
2. After each R-FCH and R-SCH frame decoding: 29
垐 ( ) ( )tx txQ Q Data FCH Data SCH (Q.1.3.2-2) 30
where ( )txData FCH and ( )txData SCH is the data transmitted in the last R-FCH 31
and R-SCH frame respectively after discounting the physical layer overhead and RLP 32
layer zero bits padding. Since RLP retransmissions are not modeled, data 33
transmitted in the last frames is discounted even if the frame is decoded in error. 34
3. At the scheduling instant it , scheduler estimates the maximum scheduled rate for 35
the MS: 36
3GPP2 C.R1002-A v1.0
202
垐( ) ( 9600) / 20 20ms
(( _ _ _ _ ) ( / 20 )
assigned
Assigned
Q f Q R ActionTimeDelay
PL FCH OHD SCH PL SCH OHD ActionTimeDelay
1
(Q.1.3.2-3) 2
3
max
max
307.2
( ),
ˆ( ) min arg max{ | ( ) (( 9600) 20ms _ _
_ _ ) ( _ / 20ms) }
RR kbps
R power
R s R Q f R PL FCH OHD
PL SCH OHD SCH PRD
(Q.1.3.2-4) 4
where SCHAssigned is an indicator function for the current scheduling period, 5
assigned
assigned
1 if R >0
0 if R = 0AssignedSCH
(Q.1.3.2-5) 6
assignedR is the rate assigned on the R-SCH during the current scheduling period and 7
MS is supposed to transmit on the R-SCH until the ActionTime of the next 8
assignment. )(max powerR is the maximum rate that the MS can support given its 9
power limit. This max rate is reported in the last received SCRM/SCRMM message. 10
11
Capacity Computation: 12
The sector capacity at the jth sector is estimated from the measured MSs‘ Sinrs. The Sinr 13
is the average pilot-weighted combined Sinr per antenna. The combining per power-14
control group (PCG) is the pilot-weighted combining over multiple fingers and different 15
antennas of the sector of interest. The combining is not over different sectors in case of 16
softer-handoff MS. The averaging is over the duration of a 20-ms frame. The following 17
formula is used for estimating Load contribution to a sector antenna [30]: 18
)( ])[,(1
])[,(
iActiveSetj FCHij
FCHij
jRERSinr
RERSinrLoad (Q.1.3.2-6) 19
where ])[,( FCHij RERSinr is the estimated Sinr if the MS is assigned a rate iR on R-SCH 20
and ][ FCHRE is the expected rate of transmission on the R-FCH. For data-only MSs with 21
autonomous transmission at a maximum rate of 9.6kbps, the maximum autonomous 22
data rate can be used for the expected value. Let the measured pilot Sinr (frame average 23
pilot Sinr averaged over two antennas) be jtcp NE )/( , while it is assigned a rate of 24
Rassign(SCH) on the R-SCH. Then, 25
( ) ( )( , ) ( / ) 1 ( / ) (( / ) ( / ))
( ( )) ( )i FCH
i FCH
j i FCH cp t j R R
assign i
Pref R Pref RSinr R R E N T P T P C P
Pref R SCH Pref R
26
(Q.1.3.2-7) 27
3GPP2 C.R1002-A v1.0
203
For voice-only MSs, the following equation is used to estimate the average received 1
Sinr: 2
max
9.6
4.8
2.7
1.5
( / )(0, [ ( )]) 1 ( 9.6 )
( ( ))
( / ) (9.6 )
( / ) (4.8 )
( / ) (2.7 )
(( / ) (1.5 )
( / )
cp t j
j FCH FCH
assign
k
k
k
k
E NSinr E R Pref R k
Pref R SCH
T P P k
T P P k
T P P k
T P P k
C P
3
(Q.1.3.2-8) 4
Scheduling Algorithm: 5
The scheduling algorithm has two fundamental characteristics: a) greedy filling for 6
maximum capacity utilization and increasing TDM gain, b) prioritization of MS rate 7
requests such that MSs which were scheduled in the last scheduling period are last in 8
the queue. This resembles Round-Robin scheduler while allowing multiple 9
transmissions for full capacity utilization. 10
Initialization: The MS rate requests are prioritized. Associated with each MS is a priority 11
count PRIORITY. PRIORITY of a MS is initialized to 0 in the beginning of the simulation. 12
When a new MS enters the system with sector j as the primary sector, its PRIORITY is 13
set equal to the min{PRIORITY , such that MS has sector as the primary sector}ii i j 14
1. Let the Load constraint be LoadLoad j max , such that the rise-over-thermal 15
overshoot above certain threshold. For the calibration purposes, max Load value of 16
0.45 will be used by the scheduler. The capacity consumed due to pilot 17
transmissions and transmissions on fundamental channels (due to voice or data) is 18
computed and the available capacity is computed as 19
ActiveSetj FCHj
FCHj
RESinr
RESinrLoadjCav
])[,0(1
])[,0(max)( (Q.1.3.2-9) 20
where max Load is the maximum Load for rise-over-thermal outage criteria is 21
satisfied. 22
MS rate requests are prioritized in decreasing order of their PRIORITY. So MSs with 23
highest PRIORITY are at the top of the queue. When multiple MSs with identical 24
PRIORITY values are at the top of the queue, the scheduler makes a equally-likely 25
random choice among these MSs. 26
2. Set k=1, 27
3. The data-only MS at the kth position in the queue is assigned the rate kR given by 28
3GPP2 C.R1002-A v1.0
204
max
( , [ ])| ( )
1 ( , [ ])min ( ),arg max
(0, [ ])0; ( )
1 (0, [ ])
j FCH
j FCHk
kR j FCH
j FCH
Sinr R E RR Cav j
Sinr R E RR R s
Sinr E Rj ActiveSet k
Sinr E R
(Q.1.3.2-10) 1
The available capacity is updated to: 2
( , [ ]) (0, [ ])( ) ( ) ; ( )
1 ( , [ ]) 1 (0, [ ])
j k FCH j FCH
j k FCH j FCH
Sinr R E R Sinr E RCav j Cav j j ActiveSet k
Sinr R E R Sinr E R
3
(Q.1.3.2-11) 4
4. If 0)(max sRk and 0kR , increment PRIORITY of the MS 5
Otherwise, do not change PRIORITY of the MS 6
5. k = k+1; if k < total number of MSs in the list, Go to Step 3, otherwise, stop. 7
Q.2 Baseline specific simulation parameters 8
9
Parameter Value Comments
R-FCH frame format 20 ms, RC 3
(9.6, 4.8, 2.7, 1.5) kbps
Specified in [29]
Voice-only MS uses all four rates
Data-only MS uses only 1.5 kbps
and 9.6 kbps
R-SCH frame format
20 ms, RC 3
(9.6 k convolutional)
(19.2, 38.4, 76.8, 153.6,
307.2) kbps turbo
Specified in [29]
SCRMM Overhead One 5-ms R-FCH
transmission 24 bits of upper layer payload
SCRM Overhead One 20-ms R-FCH
transmission
Minimum of 144 bits of upper
layer payload
Headroom_Req 5 dB
Conservative rate request
Keeps power headroom for long-
term channel variation
Reduces DTX on R-SCH
Headroom_Tx 2 dB
Reduces probability of power
outage during the duration of R-
SCH transmission
3GPP2 C.R1002-A v1.0
205
Parameter Value Comments
Average Tx Power Filter
Coefficient Headroom 1/16
Normalized Average transmit
pilot power is computed as
filtered version over several PCGs
Encode_Delay 2.5 ms
Delay between encoding of
transmission packet and actual
transmission
SCRMM Delay (Method
a)
D_RL(request)
Mean 23.625 ms
Std Dev. 6.85 ms
Uniform [11.75, 35.5] ms
Modeled as a random variable in
Appendix R
SCRM Delay (Method b)
D_RL(request)
Mean 38.625 ms
Std Dev. 6.85 ms
Uniform [26.75, 50.5]
Modeled as a random variable in
Appendix R
ESCAMM Delay (Method
a) D_FL(assign)
Mean 22.375 ms
Std dev. 3.25 ms
Uniform [16.75, 28]
Modeled as a random variable in
Appendix R
ESCAM Delay
(transmission on F-
PDCH) D_FL(assign)
X + Y
X: Uniform [11.75, 24.25]
Y: LogNormal with mean and
std. Dev. as specified in
Table R.5, Appendix R.
Modeled as a random variable in
Appendix R
ActionTimeDelay
(Method a) 31.25 ms
Based on the ESCAMM delay,
including the 2 PCG MS
encoding delay
ActionTimeDelay
(Method b) 77.5 ms
Based on the ESCAM delay on F-
PDCH at the primary sector
Geometry of –5 dB. This
includes the 2 PCG MS encoding
delay
SCH_PRD 200 ms
Scheduling Period sufficiently
long to reduce overhead while
small enough for effective
scheduling. The assigned rate
for a given MS does not have to
be over the whole 200 ms.
BUF_DEPTH (19200 X SCH_PRD) = 3840
bits
Minimum buffer size for
SCRMM/SCRM to be sent
PL_FCH_OHD 24 bits 24 bits per frame is the physical
layer overhead on the R-FCH
3GPP2 C.R1002-A v1.0
206
Parameter Value Comments
PL_SCH_OHD 24 bits 24 bits per frame is the physical
layer overhead on the R-SCH
Geometry 0.05
Constant used in the
computation of geometry for F-
PDCH delay in Appendix R
CQICH-to-(normalized)
pilot ratio (C/P)
-5.4 dB (Active Set includes
sectors from single cell)
-4.3 dB (if Active Set
includes sectors from more
than one cell)
Overhead transmission on RL
due to CQICH is modeled as a
continuous transmission with
fixed C/P ratio. For a MS not in
SHO, average C/P value
corresponding to one full CQICH
burst and 16 differential bursts
is used. For the MS in SHO,
average CQICH value
corresponding to two full bursts
and 14 differential bursts is used
Pref(Data Rate) Pilot Reference Level [dB]
Pilot Reference Levels are
determined such that an average
FER of 1% is observed across
different channel models based
on Evaluation Methodology
approved by WG3
9.6kbps 0.0
19.2kbps (turbo) 1.25
38.4kbps (turbo) 2.5
76.8kbps (turbo) 3.75
153.6kbps (turbo) 5.625
307.2kbps (turbo) 8.25
1
3GPP2 C.R1002-A v1.0
207
APPENDIX R: MODELING OF D_RL(REQUEST) AND D_FL(ASSIGN) 1
The four tables below give the minimum expected delays and worst-case delays with 2
SCRMM/SCRM transmission for request and ESCAMM/ESCAM transmissions for 3
assignment. 4
Table R-1: D_RL(request) delay for Method a 5
Minimum
Expected
delay (ms)
Cumulative
Expected
Delay (ms)
Reasonable
Worst Case
Delay (ms)
Cumulative
Reasonable
Worst Case
Delay (ms)
MS processing request message 1 1 1 1
Request message transmission
time
5 6 5 6
Decoding time at BTS 2 8 12.5 18.5
Transmission time to BSC 2.75 10.75 8 26.5
BSC processing time 0.5 11.25 2.5 29
BSC to scheduler 0.5 11.75 0.5 29.5
Potential inter-BSC handoff 0 11.75 6 35.5
Request message framing time (time when the new data arrives in the system to when the 6
request is actually sent) is included in the table as ―Request message transmission time‖. 7
Assuming uniformly distributed backhaul delay between BTS and BSC, D_RL(request) with 8
SCRMM transmission and based on above values can be modeled as uniformly distributed 9
random variable between [11.75, 35.5] ms. 10
In calculating the transmission time to the BSC, we assumed that the backhaul bandwidth 11
on the forward link and the reverse link are identical. In order to support 2.4 Mbps forward 12
link, the backhaul would need to be at least 2 x 1.544 Mbps. If various sectors are 13
aggregated into the same backhaul, then the rate will be even greater. If an ATM cell 14
consisting of 53 octets is used to transmit the uplink request, then the transmission time is 15
B/R+D/c, where B is the number of bits to be transmitted, R is the transmission rate, D is 16
the distance, and c is the velocity of transmission in the medium. Assuming 10 km 17
distance, D, and a propagation velocity in the medium of one-half the speed of light, then 18
the transmission delay is about 0.275 ms. The actual estimate needs to take into account 19
queuing delay. If the link utilization is 0.9 (a high utilization), then the mean transmission 20
delay will be about 2.75 ms. 21
3GPP2 C.R1002-A v1.0
208
Table R-2: D_RL(request) delay for Method b 1
Minimum
Expected
delay
Cumulative
Expected
Delay
Reasonable
Worst Case
Delay
Cumulative
Reasonable
Worst Case
Delay
MS processing request message 1 1 1 1
Request message transmission time 20 21 20 21
Demodulation time at BTS 2 23 12.5 33.5
Transmission time to BSC 2.75 25.75 8 41.5
BSC processing time 0.5 26.25 2.5 44
BSC to scheduler 0.5 26.75 0.5 44.5
Potential inter-BSC hand-off 0 26.75 6 50.5
Assuming uniformly distributed backhaul delay between BTS and BSC, D_RL(request) with 2
SCRMM transmission and based on above values can be modeled as uniformly distributed 3
random variable between [26.75, 50.5] ms. 4
Table R-3: D_FL(assign) delay for Method a 5
Minimum
Expected
delay
Cumulative
Expected
Delay
Reasonable
Worst Case
Delay
Cumulative
Reasonable
Worst Case
Delay
Scheduling time 2 2 2 2
Scheduler to BSC 0.5 2.5 0.5 2.5
Transmission time to BTS 5 7.5 10 12.5
BTS processing time 3 10.5 3 15.5
Delay to transmission slot 0 10.5 5 20.5
Transmission time 5 15.5 5 25.5
Grant processing time in MS 1.25 16.75 2.5 28
Assuming uniformly distributed backhaul delay and delay to transmission slot at BTS, 6
D_FL(assign) with ESCAMM transmission and based on above values can be modeled as 7
uniformly distributed random variable between [16.75, 28] ms. Please note that it does not 8
include the delay when the R-SCH transmission is framed by the mobile to when R-SCH 9
starts transmission at the R-SCH frame boundary. This latter delay is naturally modeled by 10
the simulator. 11
A uniformly distributed random variable between [u1, u2] has a mean of (u1+u2)/2 and 12
variance of (u2 – u1)2/12. 13
Modeling downlink assignment delay of ESCAM transmission on PDCH requires modeling 14
of backhaul delays and PDCH scheduling delays. Backhaul delays as above are modeled 15
3GPP2 C.R1002-A v1.0
209
uniform. Assuming no scheduling delays, following table lists the expected minimum delay 1
and worst-case delay. 2
Table R-4: D_FL(assign) delay for Method b (excluding F-PDCH scheduling delay) 3
Minimum
Expected
delay
Cumulative
Expected
Delay
Reasonable
Worst Case
Delay
Cumulative
Reasonable
Worst Case
Delay
Scheduling time 2 2 2 2
Scheduler to BSC 0.5 2.5 0.5 2.5
Transmission time to BTS 5 7.5 15 17.5
BTS processing time 3 10.5 3 20.5
Delay to transmission slot 0 10.5 1.25 21.75
Transmission time 0 10.5 0 21.75
Grant processing time in MS 1.25 11.75 2.5 24.25
To model transmission time or the scheduling delay on the F-PDCH, we did the following 4
experiment. Assuming four users are simultaneously scheduled every 200 ms, 4 ESCAM 5
packets are produced every 200 ms. Other users are added with normal traffic to generate 6
full load on forward link. Signaling messages are given priority over data. A probability 7
density function (PDF) of three MSs at different geometries is shown in Figure R.1. For the 8
cases plotted here, we have 9
10
Mean (ms) Std Dev (ms)
16.9 16.4
19.7 18.1
42.4 90.2
3GPP2 C.R1002-A v1.0
210
Mean = 0.0169 sec., std = 0.0164 sec.
Mean = 0.0197 sec., std = 0.0181 sec.
Mean = 0.0424 sec., std = 0.0902 sec.
Transmission Delay (sec)
Transmission Delay (sec)
Transmission Delay (sec)
Mean = 0.0169 sec., std = 0.0164 sec.
Mean = 0.0197 sec., std = 0.0181 sec.
Mean = 0.0424 sec., std = 0.0902 sec.
Transmission Delay (sec)
Transmission Delay (sec)
Transmission Delay (sec)
1
Figure R-1: PDF of FL transmission delays of ESCAM on F-PDCH 2
To retain the simplicity of modeling, we recommend modeling the delays as lognormal with 3
same mean and standard deviation. 4
3GPP2 C.R1002-A v1.0
211
Figure R-2 shows some of the sample points of observed scheduling delays. 1
RL scheduling message transmission delay on F-PDCH
0.00E+00
5.00E-03
1.00E-02
1.50E-02
2.00E-02
2.50E-02
3.00E-02
-1.50E+01 -1.00E+01 -5.00E+00 0.00E+00 5.00E+00 1.00E+01 1.50E+01
Gem (dB)
Pkt
Del
(sec)
Pkt Del(s)
Expon. ( Pkt Del(s))
2
Figure R-2: ESCAM delays on F-PDCH 3
Since the signaling message is small to fit in any EP size and have priority over the data 4
transmission, downlink transmission time is pretty much independent of channel model as 5
shown in Figure R-2. We can approximate the curve above by interpolating between a few 6
points, for example, based on Table R-5: 7
Table R-5: Reference table for mean transmission times vs Geometry 8
Geometry (dB) Mean Transmission Time (ms)
-9 22
0 10
10 3
To summarize, D_FL(assign) = x + y. x is uniform [11.75, 24.25]. y is lognormal with mean 9
and standard deviation obtained by interpolating or extrapolation using the values given in 10
Table R-5 and, 11
Geometry = 1 / ( (Ioc+No)/Ior + Geometry ) 12
For geometry computation, all BTSs are assumed to be transmitting at full 13
power (20 W) 14
Ior is the total energy per chip received from the strongest sector assumed to be 15
max transmit power (20 Watts) 16
3GPP2 C.R1002-A v1.0
212
Ioc is the sum of total energy per chip from all other sectors, each assumed to 1
use max transmit power (20 Watts) 2
No is the thermal noise spectral density 3
4
5
3GPP2 C.R1002-A v1.0
213
APPENDIX S: SYMBOL SNR MODELING FOR CDM TRANSMISSION WITH RAKE 1
DEMODULATION 2
In the system level simulation, the SNR of each of Walsh decovered symbol at the output of 3
the Rake demodulator shall be computed as follows: The interference due to unrecovered 4
power (FURP) shall be modeled as an additional ray that is not demodulated by the Rake 5
demodulator. The average fractional power assigned to each of the rays recovered by the 6
Rake demodulator is given in Table 2.2.2-2. The average fractional power assigned to the 7
ray used to model FURP is also given in Table 2.2.2-2. The recovered rays and the 8
additional ray used to model the unrecovered power all fade independently of each other. 9
In the system level simulation, the long-term pilot (Ec/Io) for the j-th recovered ray from the 10
sector with index s is defined as 11
BSN
m
mTXc
jsTXc
js
c
gEN
agE
I
E
1
,0
,
,0
(S-1)
12
where 13
Ec,TX Transmit energy per chip from the base station. This is related to the 14
transmit power PTX and chip-rate W=1/Tc through the equation 15
Ec,TX = PTX × Tc = PTX/W. 16
0N The power spectral density (energy per chip) of AWGN noise 17
BSN The total number of sectors in the system 18
sg The average link gain between the subscriber and the sector indexed by s. 19
This term includes the subscriber and sector antenna gains, cable and 20
connector losses, other losses, and shadowing, but not fading. 21
aj Average fractional power of the j-th recovered ray in a particular channel 22
model of the subscriber. 23
A recovered ray from a sector may be assigned to a finger on each receive antenna of the 24
Rake demodulator only if its long-term pilot (Ec/Io) exceeds the threshold T_PATH, as 25
specified in Table 2.1.2-1. Among the eligible rays, those rays with the largest long-term 26
pilot (Ec/Io) are assigned to fingers of the Rake demodulator, up to a maximum of 27
MAX_NUM_PATHS rays. Finger assignment for a subscriber receiver is determined 28
statically, at the beginning of the simulation. A sector is said to be in the active set of the 29
subscriber if at least one ray from the sector is assigned to a finger in the Rake 30
demodulator. 31
The short-term total energy (Io)i at the i-th receive antenna of the subscriber receiver is 32
computed as follows: 33
3GPP2 C.R1002-A v1.0
214
BSN
m
mimTXcigENI
1
2
,,00
(S-2) 1
where , , ,1 , , ,[ ]T
i m i m i m J i m if sector m is in the active set of the subscriber receiver, 2
and , ,i m i m otherwise. For a sector m in the active set, J denotes the total number of 3
recovered rays used in a particular channel model of the subscriber, jmi ,, denotes the 4
samples of the fading process of the j-th recovered ray from the sector at the i-th subscriber 5
antenna, and ,i m represents samples of the fading process for the additional ray used to 6
model interference due to the FURP. For a sector m not in the active set, ,i mindicates 7
samples of the fading process between the i-th subscriber antenna and the sector indexed 8
by m, which is modeled as single path fading. 9
For a finger f associated with the j-th recovered ray from the sector with index m and the i-10
th receive antenna of the subscriber, the short-term signal energy (Ec)f is computed as 11
follows: 12
2
,,, jsisTXcfc gEE (S-3) 13
The short term total energy of the finger is defined as I0,f = (Io)i. The noise plus interference 14
energy associated with the finger is given by 15
ffcfft II
CEIN ,0
1
max
,,0,
(S-4) 16
where
maxIC
denotes the achievable C/I for the Rake demodulator, as specified in Table 17
2.1.2-1 and Table 2.2.2-2. 18
Let NF denote the total number of assigned fingers in the Rake demodulator. The signal 19
from the assigned fingers of the Rake demodulator are subjected to pilot weighted 20
combining. The resulting chip SNR after pilot weighted combining is given by 21
F
F
N
f
ft
f
fc
N
f
fc
combN
P
IE
E
I
C
1
,
,0
,
2
1
,
(S-5) 22
where P denotes the pilot filter processing gain. If the pilot signal from adjacent pilot bursts 23
are linearly interpolated in the ratio c0 and c1, where c0 + c1 = 1, then2
1
2
0
96
ccP
. 24
The average Eb/Nt per slot is computed using 25
3GPP2 C.R1002-A v1.0
215
combt
b
I
C
M
K
N
E
(S-6) 1
where K is the number of data chips per slot and M is the number of information bits per 2
packet. 3
4
5
3GPP2 C.R1002-A v1.0
216
APPENDIX T: EQUIVALENT SNR APPROACH FOR OFDM TRANSMISSION AND 1
DEMODULATION 2
The equivalent SNR of the packet is computed over a transmission period and mapped to 3
an PER using the AWGN reference curve. The AWGN reference curves are obtained by 4
simulating the turbo code performance on an AWGN channel for each transmission format, 5
defined by payload size and number of slots. The equivalent SNR of each packet is 6
calculated according to the following steps. 7
T.1 Coherence Loss due to Doppler 8
The coherence loss factor Lc accounts for SNR degradation in the presence of Doppler 9
spread, due to channel variation within an OFDM symbol. The SNR on the i-antenna and k-10
th tone during the n-th OFDM symbol, resulting from coherence loss can be modeled as 11
],,[],,[1 nkiSNRLnkiSNR c (T-1)
12
The coherence loss factor Lc due to Doppler [35] is given by 13
12
21
2NTf
L dc
(T-2)
14
where df is the Doppler frequency and NT is the OFDM symbol duration. 15
T.2 Inter-tone Interference (ITI) due to Doppler 16
The inter-tone interference term accounts for non-orthogonality of adjacent tones due to 17
Doppler and can be included as follows 18
ITI
HN
nkiSNREnkiSNR1
],,,[],,[ 12
(T-3)
19
where yx
yx
yxyxEH
11
1, denotes the harmonic sum of x and y. 20
The inter-tone interference noise floor ITIN due to Doppler [35] is given by 21
12
22
NTfN d
ITI
(T-4)
22
where df is the Doppler frequency and NT is the OFDM symbol duration. 23
T.3 Channel Estimation Loss and Pilot Weighted Combining (PWC) 24
The per-antenna SNR including the effect of imperfect channel estimation and Pilot 25
Weighted dot-product operation is given by 26
],,[],,[,],,[,],,[],,[ 223 nkiSNRnkiSNRnkiSNRnkiSNREnkiSNR PILOTPILOTH (T-5)
27
3GPP2 C.R1002-A v1.0
217
where ],,[ nkiSNRPILOT is the SNR of the channel estimate used to demodulate k-th data 1
tone during the n-th OFDM symbol. 2
The SNR of channel estimate ],,[ nkiSNRPILOT is computed as follows: Let the number of 3
pilot tones for channel estimation be P and assume these tones are used to estimate a 4
channel impulse response of length L chips, where L ≤ P. Assuming white noise, equal 5
power on data and pilot tones, and least-squares channel estimation, the SNR of the 6
channel estimate is given by 7
][
1,][],,[],,[ 2
nnnkiSNR
L
PEnkiSNR HPILOT
(T.-6)
8
where ][n is the gain from averaging the channel estimate across multiple OFDM symbols 9
and ][n models the inter-symbol coherence loss due to Doppler. The values of the 10
parameters P and L shall be specified in the system description. 11
Suppose the channel estimate used to demodulate the data symbols in the n-th OFDM 12
symbol is obtained by combining/averaging pilot tones from adjacent OFDM symbols 13
(blocks). Prior to inter-symbol averaging, the pilot tones are assumed to be normalized by 14
the relative pilot gains on the adjacent OFDM symbols. If the channel estimate averaging 15
weights are nJnnn cccc ,,2,1,0 ,,,, , and the relative pilot gains are nJnnn ,,2,1,0 ,,,, the 16
averaging gain ][n and the inter-symbol coherence loss ][n are given by 17
J
j nj
nj
J
j nj
nj
c
c
n
0 ,
2
,
2
0 ,
,
][
,
2
0
,, 2exp1][
J
j
njdnj fcn
(T.-7)
18
where nj , is the time offset between the OFDM symbol (block) associated with the 19
averaging weight njc , , and n-th OFDM symbol (block) containing the data tone to be 20
demodulated. 21
T.4 Antenna Combining with Receive Diversity 22
SNRCOMB[k,n], the SNR after diversity combining across the NANT (=1 or 2) receive 23
antennas, is given by 24
ANTN
i
nki
COMB nkiSNRnkSNR 1 3
2
,,
],,[],[
1 , where
ANTN
i
nki
nkiSNR
nkiSNR
1
2
2,,
],,[
],,[
(T.-8)
25
Clearly, for NANT = 1 (single antenna receiver), the above equation simplifies to 26
SNRCOMB[k,n]= SNR3[1,k,n]. 27
28
3GPP2 C.R1002-A v1.0
218
T.5 Averaging SNR in the constrained capacity domain 1
The Equivalent SNR (also referred to as ESM), denoted equivSNR is calculated from the 2
SNRcomb[k] of all data symbols in a packet as follows 3
nk
COMBQAM
data
avg nkSNRCN
R,
16 ],[1
(T-9)
4
avgQAMequiv RCSNR 1
16
(T-10)
5
where dataN is the number of data tones in a packet, ()16QAMC is the constrained capacity 6
formula for 16QAM. The result of this computation is used to index into the AWGN 7
reference curve, to determine the erasure probability of the underlying packet. 8
The constrained capacity ()16 QAMC is defined as follows: 9
C
YY
C
NNQAM dyypypdnnpnpSNRC )(log)()(log)()( 2216
(T-11)
10
where )(npN denotes the pdf of a zero mean complex Gaussian random variable N with 11
unit variance, and )(ypY denotes the pdf of a complex-valued random variable 12
NXY , where X is a complex-random variable with SNRXE 2
, assuming each 13
value on the 16-QAM constellation with equal probability. The plot of ()16 QAMC as a 14
function of SNR (in the logarithmic domain) is provided in Figure T-1. 15
3GPP2 C.R1002-A v1.0
219
-10 -5 0 5 10 15 200
1
2
3
4
5
6
7
SNR (dB)
Capacity (
bps/H
z)
Capacity Curves
Gaussian Signaling
16 QAM
1
Figure T-1 Constrained Capacity Curve for 16-QAM 2
3
3GPP2 C.R1002-A v1.0
220
APPENDIX U: FILE FORMATS FOR VOIP AND VT MODEL 1
U.1 Source Configuration File Format 2
The following data is in ASCII text format. 3
4
AT# AT Lay# TrafficType AudOffset VidOffset 5
6
AT#: Reference AT# for reporting AT statistics (starts at 1) 7
AT Lay#: Reference # defined in AT layout file (starts at 1) 8
TrafficType: 0 = Full Buffer, 1 = VoIP, 2 = VT 9
AudOffset: Offset of starting point in audio source file, in slots (-1 for Full Buffer) 10
VidOffset: Offset of starting point in video source file, in slots (-1 for Full Buffer and VoIP) 11
12
U.2 Source File Format 13
The following data is in ASCII text format. 14
15
NumLines 16
NumSlots 17
ArrivalSlot PacketSize PacketType (NumLines lines of this format) 18
19
ArrivalSlot: In DO slots (5/3 msec) 20
PacketSize: Application packet size, in bytes 21
PacketType: For audio, always 0. For video, 0=P-frame, 1=I-frame 22
23
U.3 Per-AT Data Reporting Format for VoIP 24
The following data is to be reported in an ASCII text file, one line per AT, in the following 25
format. 26
27
AT# PhyTP PhyPER AudTP AudFER AudTailDelay 28
29
AT#: Reference AT# for reporting AT statistics, from source configuration file 30
PhyTP: Physical layer throughput (kbps) 31
3GPP2 C.R1002-A v1.0
221
PhyPER: Physical layer packet error rate 1
AudTP: Total audio bytes delivered divided by sim time (kbps) 2
AudFER: Audio packet error rate (%) 3
AudTailDelay: TargetAudFER% tail delay point (msec) 4
5
U.4 Network Statistics for VoIP 6
Over all AT‘s in the network: 7
8
Cdf of AudTailDelay 9
Cdf of AudFER 10
11
U.5 Per-AT Data Reporting Format for VT 12
The following data is to be reported in an ASCII text file, one line per AT, in the following 13
format. 14
15
AT# PhyTP PhyPER AudTP AudFER AudTailDelay VidITP VidPTP VidIFER 16
VidPFER VidTai1Delay 17
18
AT#: Reference AT# for reporting AT statistics, from source configuration file 19
PhyTP: Physical layer throughput (kbps) 20
PhyPER: Physical layer packet error rate (relate to DARQ) 21
AudTP: Total audio bytes delivered divided by sim time (kbps) 22
AudFER: Audio packet error rate (%) 23
AudTailDelay: TargetAudFER% tail delay point (msec) 24
VidITP: Total video bytes of I-frames delivered divided by sim time (kbps) 25
VidPTP: Total video bytes of P-frames delivered divided by sim time (kbps) 26
VidIFER: Video packet error rate of I-frames (%) 27
VidPFER: Video packet error rate of P-frames (%) 28
VidTailDelay: TargetVidFER% tail delay point (msec) 29
30
U.6 Network Statistics for VT 31
Over all AT‘s in the network: 32
3GPP2 C.R1002-A v1.0
222
1
Cdf of AudTailDelay 2
Cdf of VidTailDelay 3
Cdf of VidTailDelay-AudTailDelay 4
Cdf of AudFER 5
Cdf of VidIFER 6
Cdf of VidPFER 7
8
3GPP2 C.R1002-A v1.0
223
APPENDIX V: CHANNEL PARAMETERS FOR FURGE FACTOR 1
The channel model parameters in 2.2.2.2.3 are shown in this appendix. The 64 realizations 2
of the SCM urban macro environment are used for virtual decoder generation and 3
verification. 4
5
Table V-0-1 Power Levels for 64 realizations of the SCM’s 6 paths 6
Path 1 2 3 4 5 6
1 0.4699878393 0.1121509871 0.1538803462 0.1302980517 0.1291785545 0.0045042212
2 0.3323039087 0.2030574613 0.2527225796 0.0380961519 0.0965994171 0.0772204813
3 0.1320271351 0.4005296115 0.3575193246 0.0173951555 0.0278855467 0.0646432266
4 0.2541653729 0.2348525244 0.3301762407 0.0792536764 0.0874842826 0.0140679030
5 0.1389882893 0.3641602983 0.2486146336 0.1313183277 0.0874310541 0.0294873970
6 0.1260841128 0.2340698514 0.1075228014 0.1290675514 0.3630905126 0.0401651705
7 0.2180883814 0.4066938186 0.0669638060 0.1220162399 0.1769989583 0.0092387957
8 0.1113137373 0.3546804407 0.2474272259 0.2108931409 0.0429986434 0.0326868118
9 0.4803898832 0.1150346947 0.0919443311 0.0752311247 0.1612569706 0.0761429957
10 0.0299914621 0.2706598030 0.1580067293 0.2745978541 0.2047917913 0.0619523602
11 0.0786295301 0.4743348887 0.1391244947 0.0801981832 0.1146195919 0.1130933114
12 0.2781128277 0.1789618641 0.3533711116 0.1177163082 0.0553485686 0.0164893198
13 0.1808616156 0.2205919938 0.2416364183 0.0956476464 0.1369247286 0.1243375974
14 0.5398597542 0.1240807422 0.1445597984 0.1000019498 0.0404075561 0.0510901992
15 0.2278397659 0.0696610121 0.3702863534 0.0512376086 0.2045980663 0.0763771937
16 0.2207573561 0.1518646524 0.3117512926 0.1306937351 0.1116928733 0.0732400905
17 0.1578243498 0.4513858641 0.0403248128 0.0475677131 0.1934390128 0.1094582475
18 0.2444859425 0.0518628069 0.3717194291 0.0916399827 0.1712611511 0.0690306877
19 0.1708738735 0.1279988592 0.1841019586 0.1436358658 0.1265456752 0.2468437678
20 0.1695011752 0.1170682816 0.3322744493 0.2710598577 0.0596005880 0.0504956482
21 0.1303277711 0.2307857363 0.2211942399 0.2460706913 0.0449247776 0.1266967838
22 0.5678822123 0.1888138054 0.1035607216 0.0138684238 0.0887054580 0.0371693789
23 0.3362129120 0.1815273748 0.1295062398 0.2571952939 0.0730172965 0.0225408830
24 0.7987915877 0.0217713806 0.0979083852 0.0462386742 0.0233335263 0.0119564459
3GPP2 C.R1002-A v1.0
224
Path 1 2 3 4 5 6
25 0.4274814933 0.2999245485 0.0837345141 0.0634718962 0.0876607267 0.0377268212
26 0.1954590044 0.2203344918 0.1814986578 0.2654536302 0.0404080181 0.0968461977
27 0.2048115231 0.0430998122 0.0988432889 0.4097896125 0.0802230063 0.1632327569
28 0.2156311208 0.2677454738 0.0649408004 0.2853025224 0.0920073834 0.0743726993
29 0.1864403424 0.2169027956 0.1427526067 0.1014908742 0.1728783101 0.1795350711
30 0.1693889850 0.1556035730 0.4191075325 0.1176533117 0.0271592715 0.1110873263
31 0.0964569121 0.3606993837 0.0608504242 0.0645721355 0.3567869571 0.0606341873
32 0.0566926099 0.3823313949 0.1556308684 0.0491070460 0.2360863182 0.1201517625
33 0.3402930235 0.1665251835 0.2879666665 0.1215627367 0.0303086264 0.0533437634
34 0.3465119981 0.0908137109 0.0747397299 0.1508297682 0.2246673339 0.1124374590
35 0.2927359858 0.2227908417 0.1307435678 0.1269410256 0.2137136788 0.0130749004
36 0.1197822243 0.3609509946 0.0869320584 0.1696966235 0.2106724461 0.0519656531
37 0.4763027366 0.1353219023 0.0569147466 0.2621767156 0.0560982015 0.0131856975
38 0.2967955438 0.0912677526 0.2712679935 0.1839757642 0.0854665824 0.0712263636
39 0.2477459317 0.1862119010 0.4395913476 0.0580210991 0.0235355844 0.0448941363
40 0.1480668939 0.1288990076 0.3904258770 0.1143654613 0.1728561307 0.0453866295
41 0.1210382157 0.5016352485 0.1443486988 0.1528457526 0.0424814583 0.0376506262
42 0.2870481626 0.3709873998 0.1943672517 0.0928203063 0.0397846891 0.0149921906
43 0.2133615682 0.1350368455 0.4366058542 0.0633918664 0.0360348950 0.1155689707
44 0.2680853143 0.0822559506 0.1626629320 0.3588937513 0.0851988520 0.0429031998
45 0.4470116440 0.0709692323 0.1656082643 0.0392227074 0.0836007375 0.1935874145
46 0.2708413518 0.3383716076 0.0518980861 0.1988818315 0.0511576998 0.0888494233
47 0.1241486025 0.3520581625 0.1884504166 0.0366455197 0.0720766161 0.2266206826
48 0.4354902948 0.0985662839 0.2791192387 0.0496850302 0.1168705438 0.0202686087
49 0.1683573476 0.2693203809 0.0344124950 0.0774356537 0.4500409069 0.0004332159
50 0.1809038379 0.1210333637 0.1453947318 0.3477919701 0.1086934116 0.0961826849
51 0.3170777077 0.1653350352 0.3410680596 0.0707487716 0.0337340528 0.0720363730
52 0.2866223958 0.1555833656 0.2546130487 0.1375081140 0.0240877320 0.1415853438
53 0.2812320145 0.2182422300 0.0941003415 0.0700237282 0.1365316827 0.1998700031
54 0.0461019662 0.1786750223 0.0473745037 0.5200044498 0.0818971459 0.1259469121
3GPP2 C.R1002-A v1.0
225
Path 1 2 3 4 5 6
55 0.4045866497 0.2496335144 0.0747677304 0.1448685579 0.0920867612 0.0340567864
56 0.2259569449 0.2847681920 0.1019630418 0.1277708999 0.1187752102 0.1407657113
57 0.1315633896 0.4193298292 0.0641866629 0.1614136137 0.2203497388 0.0031567658
58 0.5154432475 0.0786847679 0.0590088122 0.1057410873 0.2157936588 0.0253284263
59 0.1660378899 0.1511954089 0.3152619699 0.1786887923 0.0565884408 0.1322274982
60 0.3143491951 0.2337164259 0.2008582947 0.0685404156 0.1283614759 0.0541741928
61 0.2684282612 0.1991882420 0.1567112662 0.1115467494 0.0568304690 0.2072950122
62 0.2406470499 0.3147909629 0.1813148911 0.1778623879 0.0823385949 0.0030461134
63 0.3699285620 0.3338681556 0.1048521281 0.0871040030 0.0671645882 0.0370825632
64 0.0586134713 0.0749042496 0.3099683574 0.1302599674 0.1288477895 0.2974061648
1
2
Table V-0-2 Delays for 64 realizations of the SCM’s 6 paths 3
1 2 3 4 5 6
0.000E+00 5.696E-07 7.272E-07 1.224E-06 1.452E-06 1.999E-06
0.000E+00 2.062E-07 6.871E-07 8.734E-07 1.649E-06 1.908E-06
0.000E+00 2.425E-07 2.941E-07 2.118E-06 2.188E-06 2.199E-06
0.000E+00 6.865E-08 8.949E-08 2.320E-06 2.423E-06 2.939E-06
0.000E+00 1.623E-07 3.201E-07 5.896E-07 6.031E-07 1.311E-06
0.000E+00 5.624E-07 6.857E-07 1.101E-06 1.376E-06 3.052E-06
0.000E+00 3.380E-07 7.101E-07 9.606E-07 9.837E-07 3.862E-06
0.000E+00 9.124E-07 1.168E-06 1.218E-06 1.987E-06 3.078E-06
0.000E+00 1.447E-06 1.781E-06 1.855E-06 2.093E-06 2.576E-06
0.000E+00 1.121E-07 2.015E-07 6.737E-07 1.133E-06 1.618E-06
0.000E+00 1.890E-07 1.999E-07 1.166E-06 1.757E-06 1.851E-06
0.000E+00 1.311E-07 5.342E-07 1.166E-06 2.458E-06 4.832E-06
0.000E+00 2.232E-08 1.931E-07 9.588E-07 1.922E-06 2.540E-06
0.000E+00 7.063E-09 4.403E-08 6.363E-07 1.375E-06 2.282E-06
0.000E+00 2.770E-07 5.326E-07 8.379E-07 1.106E-06 3.327E-06
0.000E+00 6.622E-08 2.448E-07 2.725E-07 4.861E-07 2.141E-06
3GPP2 C.R1002-A v1.0
226
1 2 3 4 5 6
0.000E+00 4.773E-07 5.291E-07 6.904E-07 1.331E-06 1.759E-06
0.000E+00 4.821E-07 5.694E-07 5.779E-07 5.999E-07 3.449E-06
0.000E+00 1.444E-07 2.971E-07 3.315E-07 4.505E-07 6.295E-07
0.000E+00 4.516E-08 1.782E-07 2.008E-07 4.133E-07 1.698E-06
0.000E+00 1.742E-07 2.150E-07 4.342E-07 1.451E-06 1.986E-06
0.000E+00 2.106E-07 4.404E-07 1.221E-06 1.251E-06 1.773E-06
0.000E+00 1.894E-07 3.340E-07 8.583E-07 1.300E-06 1.488E-06
0.000E+00 5.779E-08 4.064E-07 7.493E-07 1.113E-06 1.845E-06
0.000E+00 1.218E-07 8.781E-07 1.368E-06 1.999E-06 2.426E-06
0.000E+00 1.454E-06 1.976E-06 2.144E-06 3.265E-06 5.520E-06
0.000E+00 5.039E-07 5.141E-07 8.520E-07 1.153E-06 2.121E-06
0.000E+00 3.135E-07 1.079E-06 1.179E-06 1.878E-06 4.332E-06
0.000E+00 3.079E-08 1.161E-07 1.868E-07 2.015E-07 2.513E-07
0.000E+00 5.792E-08 3.746E-07 5.723E-07 9.468E-07 1.099E-06
0.000E+00 3.248E-08 1.842E-07 7.783E-07 1.052E-06 2.058E-06
0.000E+00 4.904E-07 6.154E-07 6.194E-07 8.056E-07 2.735E-06
0.000E+00 1.682E-07 5.573E-07 1.539E-06 2.295E-06 3.478E-06
0.000E+00 4.718E-08 9.490E-08 1.270E-07 5.428E-07 6.485E-07
0.000E+00 9.609E-08 2.845E-07 3.913E-07 4.445E-07 1.278E-06
0.000E+00 1.092E-07 1.726E-07 1.546E-06 2.375E-06 2.440E-06
0.000E+00 4.112E-08 9.695E-07 4.009E-06 5.784E-06 9.494E-06
0.000E+00 6.088E-07 7.400E-07 1.272E-06 2.731E-06 3.026E-06
0.000E+00 5.990E-07 1.088E-06 1.724E-06 1.802E-06 1.904E-06
0.000E+00 3.719E-09 2.289E-08 4.605E-07 5.572E-07 9.645E-07
0.000E+00 2.704E-07 3.348E-07 5.912E-07 7.780E-07 2.721E-06
0.000E+00 2.979E-07 5.382E-07 5.860E-07 1.113E-06 5.813E-06
0.000E+00 1.147E-08 1.247E-07 2.062E-06 2.400E-06 2.433E-06
0.000E+00 3.395E-08 4.771E-08 9.985E-08 3.100E-07 1.016E-06
0.000E+00 2.104E-07 2.194E-07 2.953E-07 6.775E-07 1.082E-06
0.000E+00 2.125E-07 1.909E-06 2.439E-06 2.813E-06 3.052E-06
3GPP2 C.R1002-A v1.0
227
1 2 3 4 5 6
0.000E+00 6.272E-09 4.048E-07 9.895E-07 1.110E-06 1.379E-06
0.000E+00 4.056E-07 5.293E-07 1.050E-06 1.055E-06 3.910E-06
0.000E+00 1.155E-07 3.271E-07 3.815E-07 4.867E-07 2.192E-06
0.000E+00 1.778E-07 5.887E-07 6.751E-07 8.898E-07 1.396E-06
0.000E+00 4.178E-08 6.851E-08 6.329E-07 6.466E-07 2.882E-06
0.000E+00 4.094E-07 8.066E-07 1.582E-06 1.959E-06 2.780E-06
0.000E+00 7.852E-08 1.885E-07 5.617E-07 6.155E-07 7.129E-07
0.000E+00 3.333E-07 4.673E-07 5.069E-07 2.225E-06 2.834E-06
0.000E+00 2.803E-07 6.344E-07 9.027E-07 9.208E-07 3.131E-06
0.000E+00 3.057E-07 6.091E-07 7.369E-07 9.518E-07 1.318E-06
0.000E+00 4.799E-08 2.462E-07 3.365E-07 5.288E-07 3.063E-06
0.000E+00 6.872E-07 7.733E-07 1.048E-06 1.151E-06 3.666E-06
0.000E+00 1.383E-07 4.949E-07 5.170E-07 5.746E-07 6.565E-07
0.000E+00 1.149E-08 9.584E-08 4.884E-07 8.514E-07 2.028E-06
0.000E+00 5.433E-08 5.519E-08 2.315E-07 2.773E-07 2.981E-07
0.000E+00 6.553E-07 8.700E-07 1.925E-06 3.667E-06 9.838E-06
0.000E+00 1.515E-08 8.346E-07 3.711E-06 4.371E-06 7.680E-06
0.000E+00 3.473E-07 4.626E-07 6.514E-07 8.998E-07 1.320E-06
1
2
Table V-0-3 AoD for 64 realizations of the SCM’s 6 paths 3
1 2 3 4 5 6
6.19 -9.84 -14.34 18.63 32.76 -41.09
4.91 -5.75 -11.40 -18.60 20.97 31.11
0.06 6.17 12.56 -15.44 -19.10 -51.91
-5.35 8.29 11.28 17.66 18.67 23.25
-1.64 -2.54 4.24 -5.36 17.18 -21.33
-7.65 -12.97 -13.59 -21.76 -36.45 -54.38
-4.10 4.74 -6.89 -18.34 20.46 34.09
2.64 -16.43 -29.83 -29.99 -44.48 -54.91
3GPP2 C.R1002-A v1.0
228
1 2 3 4 5 6
-0.26 -2.16 -5.05 5.64 -8.25 -18.01
4.49 -8.25 14.86 29.90 -37.75 57.13
-0.09 5.36 14.04 15.83 -18.45 -23.77
-4.32 -11.63 -12.03 -13.37 -13.81 17.13
-12.96 13.52 14.45 -21.82 34.71 -39.94
-0.45 -3.10 7.33 -11.17 -13.95 26.65
-0.48 -1.69 -8.66 -8.86 -18.20 27.94
0.67 25.44 -28.61 -44.64 66.44 -70.12
-1.80 -12.89 19.94 -25.72 32.69 45.11
4.90 -8.22 -8.58 16.92 -21.33 -31.56
-0.50 -6.05 9.72 11.59 -13.27 -17.70
3.93 -5.37 -9.00 -12.59 15.94 38.14
-8.45 -12.70 -34.00 -39.79 54.76 73.91
2.42 11.96 -17.40 -20.55 25.58 29.18
-0.23 8.02 -8.22 -17.57 26.11 -64.32
-4.08 -8.47 -14.59 -18.61 -26.33 -30.66
7.27 -12.31 -15.01 19.40 19.70 -44.79
-1.07 6.09 7.61 -12.41 17.00 -28.18
-7.79 9.35 10.11 10.11 -11.37 16.91
-4.81 -8.60 15.30 21.42 -21.55 -41.41
-0.06 -0.86 -4.41 -5.09 10.20 19.42
-1.33 -1.46 12.21 13.31 16.37 16.56
-5.04 6.18 -13.44 16.21 -21.60 33.99
1.61 -5.27 -6.82 -12.47 -12.56 17.90
-2.35 -18.18 27.92 -29.45 34.97 60.69
-21.43 22.07 -26.60 31.05 32.26 -41.47
2.01 -5.07 -5.78 -6.41 8.94 -12.11
6.94 7.50 9.51 -10.66 -15.04 -19.19
1.29 2.49 -3.71 -12.09 22.66 -35.28
1.04 7.69 -21.19 -23.87 -29.64 -120.95
3GPP2 C.R1002-A v1.0
229
1 2 3 4 5 6
-3.01 -7.70 -11.89 12.22 -15.90 -36.00
3.58 -5.79 10.76 -26.41 -30.77 -71.23
-2.74 9.77 -12.94 19.90 -25.48 -39.48
2.05 8.45 8.53 -9.19 -20.19 -29.66
7.30 -11.30 -26.94 -38.78 44.27 -50.27
3.88 -5.20 -8.35 23.99 -29.13 -45.78
1.51 4.60 -8.18 -15.77 -20.76 -21.87
4.53 -7.01 9.72 11.03 19.96 -49.49
2.64 13.88 18.59 -25.51 -30.00 -35.12
-4.46 -4.52 6.86 -8.57 17.15 -41.03
3.42 -11.49 -15.88 -18.12 -27.50 49.88
-10.76 23.29 38.41 42.29 -72.44 -93.11
4.23 -6.14 12.52 -19.05 -23.00 23.16
7.02 -8.15 10.74 -18.19 24.02 -27.88
-12.84 13.09 -24.32 28.74 29.01 -39.62
0.40 17.78 27.93 -33.15 40.15 -62.98
1.35 -7.78 10.73 -13.58 -16.05 33.73
12.17 -13.17 -17.25 -18.61 -37.33 -51.22
-1.72 -3.51 -4.24 -4.95 -9.96 17.34
-4.25 6.43 -12.94 14.66 -20.33 22.33
-7.01 -11.12 15.20 -15.93 17.70 44.60
3.55 7.26 -8.96 -41.75 43.35 51.47
2.34 15.46 -15.79 -17.47 -26.57 34.67
-1.92 8.82 -9.49 9.68 -14.11 28.79
21.07 28.75 44.15 47.60 -83.31 -85.54
5.37 8.90 10.31 -14.47 16.29 25.97
1
Table V-0-4 AoA for 64 realizations of the SCM’s 6 paths 2
1 2 3 4 5 6
-35.19 34.83 -72.51 -13.41 98.08 -77.68
3GPP2 C.R1002-A v1.0
230
1 2 3 4 5 6
60.88 4.97 -10.22 -128.09 -106.43 54.46
46.04 -11.96 6.44 9.49 104.50 45.00
-125.44 -97.48 -40.51 36.89 -53.70 142.51
-81.58 3.03 14.63 58.32 7.32 123.45
-135.22 -15.28 -122.17 -40.91 18.92 45.19
37.29 39.16 75.55 -126.32 -26.01 14.39
27.55 -38.45 80.34 -55.07 139.24 90.05
-50.85 12.99 10.91 23.17 -89.10 -259.47
66.13 -78.10 109.38 21.86 30.82 87.56
-53.37 -100.16 -69.75 -160.30 -139.60 -35.37
-3.51 -53.26 -92.57 1.94 -12.88 -149.70
3.03 113.50 85.32 -74.88 -21.08 2.99
-129.30 -7.20 8.32 -113.43 -81.16 -37.28
-42.95 -37.94 -43.63 7.01 99.59 -165.09
12.48 -68.66 71.70 36.19 -87.37 3.27
-120.02 35.23 -79.61 29.68 -19.29 -132.29
-89.28 -104.36 22.89 95.26 -53.65 -31.35
71.56 46.31 144.12 -55.42 47.69 -94.22
125.67 -6.18 -66.94 113.51 -38.17 -109.41
-41.61 -53.32 74.00 21.06 28.39 137.68
11.24 15.76 28.01 -112.03 49.05 97.91
69.08 -54.16 -47.40 142.40 -92.60 42.62
-57.84 -43.51 -15.27 -24.08 -88.05 24.14
78.77 61.53 159.88 -146.03 -160.58 -9.04
-54.56 -98.20 31.38 -84.78 -134.71 -48.86
55.62 -201.85 -32.51 -13.95 -84.31 -48.48
-21.58 -69.87 131.48 36.09 24.27 -122.91
-32.84 -80.67 41.24 48.75 192.37 -98.79
87.25 -15.22 -15.29 116.45 177.17 44.18
45.65 -154.17 31.56 68.75 125.64 96.25
3GPP2 C.R1002-A v1.0
231
1 2 3 4 5 6
187.97 11.98 133.04 -97.97 -52.72 -44.79
126.29 36.22 -66.73 -2.14 -98.80 31.30
1.22 103.14 27.44 57.81 7.51 76.79
-46.74 85.55 87.33 41.95 9.92 26.88
9.91 -18.20 20.92 96.67 -31.67 55.27
-51.55 -71.46 -128.75 -114.61 53.94 24.05
-88.62 13.56 163.59 -65.86 113.03 75.93
-54.28 79.78 -16.24 -40.62 -175.52 73.06
51.37 56.91 67.19 299.97 58.22 -74.30
-34.96 61.66 9.51 59.72 -36.18 185.55
-2.05 -28.56 91.88 -19.41 35.35 0.78
-33.72 -39.01 -127.20 105.00 96.27 -77.82
4.21 -242.17 -61.89 75.28 -18.86 133.00
3.62 -47.60 -60.50 -58.47 86.79 29.37
35.47 -5.51 47.67 78.42 84.49 -26.38
50.02 28.72 -92.57 -28.70 93.80 100.58
17.81 -46.60 -6.18 72.14 -142.77 60.97
-79.00 51.38 142.28 68.65 40.38 -90.21
2.89 -46.65 90.98 105.31 32.51 112.18
32.63 -154.85 18.13 -51.03 -84.96 135.21
-88.84 55.11 -39.94 -83.47 59.04 -6.38
-90.46 1.00 18.78 56.81 16.65 29.44
-70.05 21.67 165.06 15.54 81.19 21.88
2.79 -17.34 -29.92 53.15 -85.90 -103.63
161.00 -54.54 2.17 3.72 76.00 -150.85
173.68 -132.87 115.33 39.29 -44.00 91.70
-59.69 79.77 -10.45 175.78 -83.13 -46.71
-41.11 16.21 -6.04 -3.13 -233.21 -56.99
5.01 -49.92 -46.36 -33.07 71.02 13.40
-18.76 76.95 -46.01 144.10 -28.36 -65.31
3GPP2 C.R1002-A v1.0
232
1 2 3 4 5 6
-122.52 -74.05 -6.62 126.60 -135.19 54.53
-12.09 97.94 -6.13 18.53 30.03 -25.93
-44.05 -210.36 -18.58 -140.39 45.94 -13.19
1
2
Table V-0-5 AT orientation relative to array normal and complex velocity vector angle 3
Array Orientation
Broadside from the LOS
direction (deg)
Direction of Travel
measured from the Array
Broadside (deg)
342.92 42.30
266.72 126.78
155.31 223.90
153.00 270.90
242.12 295.02
210.88 316.32
356.88 190.00
16.31 174.74
289.67 204.97
122.74 250.75
4.92 117.33
259.16 272.43
66.53 314.21
55.49 186.70
355.49 182.52
56.02 254.54
10.49 184.92
308.95 60.19
182.83 20.75
186.48 260.27
247.54 134.43
3GPP2 C.R1002-A v1.0
233
Array Orientation
Broadside from the LOS
direction (deg)
Direction of Travel
measured from the Array
Broadside (deg)
25.55 166.19
310.07 280.61
274.31 355.58
141.33 274.73
312.72 149.95
257.12 141.04
231.14 98.05
171.04 37.27
351.89 244.64
5.03 67.67
40.33 284.89
262.51 320.57
211.07 16.88
24.57 106.37
330.67 176.47
269.84 44.25
286.25 137.25
302.66 261.72
49.52 137.32
342.90 125.68
212.03 334.80
9.28 146.66
295.92 171.86
268.87 34.11
256.97 123.21
168.06 191.82
332.39 149.42
80.25 197.42
3GPP2 C.R1002-A v1.0
234
Array Orientation
Broadside from the LOS
direction (deg)
Direction of Travel
measured from the Array
Broadside (deg)
299.22 312.74
214.60 121.01
305.58 353.68
52.52 262.73
119.05 71.58
164.64 206.67
68.56 149.90
185.60 145.88
233.82 144.43
248.23 355.88
272.48 108.78
2.41 171.30
358.49 337.43
354.20 242.04
148.74 144.94
1
2
3GPP2 C.R1002-A v1.0
235
APPENDIX W: LINK LEVEL STATISTICS FOR GENERATING THE SHORT-TERM FER 1
CURVES FOR LINK-TO-SYSTEM MAPPING 2
This section describes the link level statistics that are used for generating the short-term 3
packet error rate (PER) curves for link-to-system mapping. The average PER and SINR 4
definitions that can be used for reporting link efficiency are also described. 5
W.1 Terminology 6
Table W-0-1 Terminology for Link Level Statistics 7
Notation Description
PHY PDU (PPDU) A physical layer protocol data unit (packet)
that is injected into the physical layer from
the layer above.
Sub-frame The unit of time of the PHY layer
operation.
)(k Channel SINR measured during the k-th
sub-frame. This includes any overhead
due to pilot, cyclic prefix etc.
)(laccum Accumulated traffic SINR for the l-th
transmission of a PPDU (with 0)( laccum if
the the PPDU did not have an l-th (re)
transmission). This SINR is without the
pilot, cyclic prefix overheads (etc.).
subframeN Number of sub-frames simulated.
ppduN Number of PHY PDUs simulated.
],...,2,1[ LN term ][lN term represents the number of PHY
PDUs that terminated at the lth
(re)transmission. L is the maximum
number of retransmissions.
],...,2,1[],,...,2,1[ LNLN failsuccess ][/][ lNlN failsuccess represent the number of
PHY PDUs that succeeded/failed at the lth
(re)transmission out of the ][lN term PHY
PDUs that terminated in the l-th
transmission. Termination with error
could either be due to ACK/NAK signaling
errors or due to reaching the maximum
limit of L transmissions.
le Error event (CRC failure) in the l-th
3GPP2 C.R1002-A v1.0
236
Notation Description
transmission.
ls CRC success, ls = le1
1
W.2 PER and SINR Definitions 2
Given the Markov model of the HARQ channel, i.e., 3
))(,|Pr())(,,...,,|Pr( 1121 leeleeee accumllaccumlll , and that for any reasonable decoder, 4
0)](,|Pr[ 1 lse accumll , the probability of error for the l-th transmission given )(laccum can 5
be written, 6
)]1(|Pr[)](,|Pr[)](|Pr[
)]1(|Pr[)](,|Pr[)]1(|Pr[)](,|Pr[)](|Pr[
11
1111
leleele
lslseleleele
accumlaccumllaccuml
accumlaccumllaccumlaccumllaccuml
7
(W-1) 8
The conditional short-term PER is defined as )](,|Pr[ 1 lee accumll . This is calculated using 9
AWGN simulations for different values of )(laccum as follows: 10
L
ln
term
l
n
successtermppdu
accumll
nN
lNnNN
lee
)(
)()(
)](,|Pr[
1
11 (W-2) 11
The unconditional short-term PER is defined as )](|Pr[ le accuml , also collected using AWGN 12
simulations. This is defined as: 13
ppdu
l
n
successtermppdu
accumlN
lNnNN
le
1
1
)()(
)](|Pr[
(W-3) 14
Either the conditional or the unconditional short-term PER can be used as the link-to-15
system short-term curves for CRC event generation. 16
The output metric of link-level simulations with fading to verify the accuracy of the link-to-17
system mapping is the average unconditional PER, defined as: 18
)]}(|{Pr[]Pr[ leEe accumll (W-4) 19
For power-controlled simulations, the above average unconditional PER after a certain 20
targeted number of transmissions is kept constant. To measure link efficiency with a 21
certain fixed PER target, we can define the average SINR per chip (Ec/Nt) as, 22
3GPP2 C.R1002-A v1.0
237
subframe
k
N
k
)(
(W-5) 1
The per-PPDU SINR, that takes into account early-termination gain, is defined as, 2
ppdu
kPPDU
N
k
)(
(W-6) 3
The average Eb/Nt needed per bit is given by, 4
effPPDUc
ppdu
kb SF
A
N
N
k
)(
(W-7) 5
6
W.3 Examples 7
The example of Figure 1 includes the transmission of 8 PPDUs that consume 14 sub-8
frames. In the figure, the arrows indicate the time number of transmissions each PDU 9
required until terminated. Note that the maximum number of transmissions, L, is set to 3, 10
and in the 3 sub-frame packet termination case, only one is positively acknowledged. 11
12
CRC=1CRC=1 CRC=1 CRC=0CRC=0 CRC=0 CRC=0 CRC=1 CRC=0 CRC=0 CRC=0 CRC=1CRC=1 CRC=1
1 2 3 4 5 6 7 8
PHY PDUs
13
Figure W-1 Example for Calculating SINR Statistics for Link Level and Virtual Decoder 14
Simulations 15
16
2/12/)]124(8[)|(;8/18/)]124(8[)(
2/1)22/()]24(8[)|(;4/18/)]24(8[)(
2/18/)48()(
);1)3(,1)3((2)3();2)2(,0)2((2)2(
);4)1(,0)1((4)1(,8,14
233
122
1
eePeP
eePeP
eP
NNNNNN
NNNNN
successfailtermsuccessfailterm
successfailtermppdusubframe
17
14
...)(
141
subframe
k
N
k
(W-8) 18
3GPP2 C.R1002-A v1.0
238
8
...)(
141
ppdu
k
PPDUN
k
(W-9) 1
2