Upload
hall-lee
View
37
Download
6
Tags:
Embed Size (px)
DESCRIPTION
Helsinki University of Technology Networking Laboratory. Impact of GPRS on existing GSM services. Juan Ventura. Supervisor: Professor Raimo Kantola Instructor: Ph. D. Peng Zhang. Table of contents. Background GPRS service - PowerPoint PPT Presentation
Citation preview
1
Impact of GPRS on existing GSM services
Juan Ventura
Supervisor: Professor Raimo KantolaInstructor: Ph. D. Peng Zhang
Helsinki University of TechnologyNetworking Laboratory
2
Table of contents
Background
GPRS service
Impact on existing GSM services
Simulation model
Simulation results
Conclusions
Future work
3
Background (I)
Impressive growth of Internet Success of mobile networks
wireless Internet
Current cellular networks
circuit switching
Highly inefficent utilization of radio resources
Service expensive time-oriented charging
Slow data rates
Inefficiencies of circuit-switched mobile networks for transporting bursty data traffic
Forburstydata
traffic
4
Packet switching techniques in wireless networks
Statistical multiplexing optimized usage of resources
Multi-slot operation higher data rates
Shorter access times
Cheap service volume based charging
General Packet Radio Service (GPRS) for GSM
Background (II)
CDPD, GPRS, etc
5
New network nodes
Changes in NSS and BSS
Impact on network planning
SGSN, GGSN, backbone network, firewalls, border gateways
GPRS register in HLR
New interfaces between “old” and “new” nodes
Protocol control unit (PCU) and channel control unit (CCU)
New resources for GPRS
GPRS - Impact on GSM network
Impact on existing GSM services Impact on existing GSM servicesSubject of the Thesis
6
Two alternatives when GPRS introduction:
1. Allocating new spectrum for GPRS
High investments in new cell sites or new TRXs
Waste of unused GSM capacity
Frequency re-plan
2. Sharing current spectrum between GPRS and GSM
Dedicating GSM traffic channels to GPRS only
Dynamic sharing between GPRS and GSM, with GSM priority
GPRS - Resource management
(I)
2. Sharing current spectrum between GPRS and GSM
three techniques
7
1. Complete Partitioning
2. Complete Sharing
3. Partial Sharing
• Always gives the best GPRS performance• Allows more flexibility in catering to the QoS requirements• Best able to adapt to a changing network load profile
PARTIAL
SHARING
Channel allocation techniques in a GPRS/GSM network:
1. Complete Partitioning
2. Complete Sharing
3. Partial Sharing
Reduction of Capacity&Quality of existing GSM services
GPRS - Resource management
(II)
8
Impact on existing GSM services (I)
Effects of GPRS partial sharing implementation on existing GSM services:
1. Interference effects - reduction of quality (QoS)
The interference probability of GSM services increases
2. Blocking effects - reduction of capacity (GoS)
Less traffic channels available for GSM services Difference between new calls and handovers
Handover performance major criterion in a GSM network
Guarantee the QoS and GoS of existing GSM services at the same time that having an effective GPRS service
Network planning problem
2. Blocking effects - reduction of capacity (GoS)
otherstudies
Thesisgoal
9
Impact on existing GSM services
(II)
Several ways of counteracting the reduction of capacity:
New frequency assignment strategies
New bandwidth for the operators
New TRXs without allocating new bandwidth
For future UMTS networks, handover between GPRS and UMTS
Queueing new call attempts
(.....)
Handover prioritization schemes
high cost, scarce spectrum
high cost, frequency re-plan
easy, unacceptable handover failure
easy, cheap, increase in call blocking
Improving GoS of existing GSM services at the same time that prioritize handovers over new calls
Handover prioritization schemes
high complexity
10
Impact on existing GSM services
(III)
Handover prioritization schemes:
1. Non-prioritize scheme (NPS)
Both new calls and handovers are handled without preference Most typically employed by cellular technologies
2. Reserved channel scheme (RCS)
Reserving a number of channels exclusively for handovers 2.1. Pre-reservation 2.2. Post-reservation
3. Queueing priority scheme (QPS)
Existence of handover area
3.1. FIFO priority queueing 3.2. Measurement-based priority queueing (MBP)
11
Impact on existing GSM services
(IV)
Handover prioritization schemes:
4. Sub-rating scheme (SRS)
Sub-rating an existing call to accommodate a handover Penalty: reduction of voice quality
5. Hybrid schemes
Combination of the aboved schemes
•Evaluate the performance degradation of GSM traffic when GPRS partial sharing implementation•Evaluate the effectiveness of these handover schemes for improving handover performance
Researchproblem
Practicalapproach
Simplified case study of a GPRS/GSM network
SIMULATION
12
Simulation (I)
Simulation methodology event-driven simulator
Simulation library developed in C++
System model:
Single cell microcell with 4 TRXs• 3 signalling channels• 29 traffic channels
Uplink procedure resource contention/reservation
Fixed channel allocation
Traffic models
Mobility models
(...)
13
Evaluation criteria:
Probability of new call blocking
Probability of handover failure
Carried traffic (network capacity)
Channel utilization
Simulation (II)
Two different scenarios for the microcell:
Basic microcell scenario
Overlaid macrocell/microcell scenario
•Fast handovers carried by the umbrella macrocell•Macrocell overlaying 7 microcells
14
Simulation results (I)
1. Basic microcell scenario:
Non-prioritized scheme (NPS):•Effects of increasing Ngprs (growth of subscriber numbers)
almost negligible unacceptable
considerable regarding Phf-handover schemes-
15
Simulation results (II)
Reserved channel scheme (RCS) Ngprs = 4
•Effects of increasing Nho
RC
S-p
reR
CS
-pos
t
16
Simulation results (III)
Queueing priority scheme (QPS) Ngprs = 4
•Two different queueing policies: FIFO and MBP•Two degradation intervals for comparison purposes
Performance of both queueing schemes is roughly the same Better performance than RCS but more implementation complexity
17
Simulation results (IV)
Sub-rating scheme (SRS) Ngprs = 4
•Effects of increasing Nsub
Best handover performance without degrading Pnb Highest implementation complexity
18
Simulation results (V)
2. Overlaid macrocell/microcell scenario:
•Ngprs = 4 •Aoff = 21Erlangs Phf = Pnb = 2% (rush hour - worst case situation)
Overall teletraffic performance is enhanced Multiple layers of cells in current GSM networks
19
Conclusions (I)
Particular results for a microcell with 4 TRXs:
For Ngprs =1, 2 (“low” GPRS penetration factor):•Capacity reduction of GSM traffic is almost negligible•Benefit of reserving additional channels to GPRS users
For Ngprs=4 (“medium” GPRS penetration factor):•Capacity reduction of GSM traffic is considerable, but can be overcome to a certain extent by using handover schemes
•An umbrella macrocell improves the overall performance
For Ngprs=6, 8 (“high” GPRS penetration factor):•Capacity reduction of GSM traffic is excessive and an umbrella macrocell is not enough to ensure sufficient capacity
•Capacity expansion is necessary (new TRXs or new cell sites)
20
Conclusions (II)
General conclusions:
Ngprs GSM capacity degradation
Depending on the type of cell and the value of Ngprs different handover schemes to be used
Selection of a particular handover prioritization scheme tradeoff between its implementation complexity and performance
•NPS, RCS When implementation cost is a major concern
•SRS Best handover performance and highest network capacity
•QPS Best choice in terms of hand. performance and implem. complexity
Significant growth of GPRS users capacity expansion compulsory
21
Future work
Performance study of both services (GSM and GPRS) using different handover prioritization schemes
Performance tradeoff between both services
Different scenarios
Changes required in the simulator:
Accurate GPRS traffic model (e.g. ETSI WWW traffic model)
More complex simulation environment
Less restrictive assumptions
more validation for the results
22
Questions?