View
299
Download
12
Category
Preview:
Citation preview
Quality of Service Aware Medium Access in Coexisting Cognitive
Radio Networks
Student:Iffat AnjumRoll: RK-554MS 3rd Semester
Thesis Supervisor:Md. Abdur RazzaqueProfessor
May 2015Department of Computer Science and Engineering
University of Dhaka
Presentation on MS Thesis Defense
2ContentsIntroductionChallenges
Design GoalsRelated WorkContributionSystem Model
Proposed WF-MACPerformance Analysis
ReferencesConclusion
3Introduction
4Introduction
[D. Goldman, “FCC scrambles to cope with data avalanche,” http://money.cnn.com/2011/12/29/technology/whitespace spectrum/index.htm , accessed on May 2015][Facilitating Opportunities for Flexible, Efficient, and Reliable Spectrum Use Employing Cognitive Radio Technologies, Federal Communications Commission (FCC), Washington, D.C. 20554, December 2003]
5Introduction• Cognitive Radio (CR)
▫ Attempts to opportunistically transmit in licensed frequencies, without affecting the pre-licensed users of these bands
▫ Its aware of its surroundings and adapts intelligently
Figu
re :
Cog
nitiv
e R
adio
Net
wor
k
Figure: Opportunistic Spectrum Usage
6Introduction
Figure : Performance Improvements Achieved by CR
7
• Full Cognitive Radios do not exist at the moment and are not likely to emerge until 2030
• Some technologies are available with some elements of CR
▫ Adaptive allocation of frequency channels in Digital Enhanced Cordless Telecommunications (DECT) wireless telephones,
▫ Adaptive power control in cellular networks, ▫ Multiple input multiple output (MIMO) techniques,▫ TV white space, etc.
Introduction
[S. Pollin, M. Timmers, and L. Van der Perre, “Emerging standards for smart radios: Enabling tomorrow’s operation,” in Software Defined Radios, 1st ed., ser. Signals and Communication Technology. Springer Netherlands, 2011, pp. 11–35.][S. D. R. F. Inc., “SDR: Software defined radio,” https://forums.hak5.org/index. php?/forum/81-sdr-software-defined-radio , accessed on March 2015]
8Introduction
Cognitive radio networks are expected to be ubiquitous and multiple CRNs often coexist with each other
Figure: Coexistence of Cognitive Radio Networks
9Motivation• When multiple CRNs operate using the same set of
channels, there is a possibility that the SUs will try to act greedily and occupy all the available channel bandwidth.
• Miss-conception on channel occupancy.▫ Starvation▫ Low throughput
• The number of CR cells from multiple CRNs typically exceeds the number of channels. ▫ Interference▫ Repeated channel switching
• Diverse user applications produces data with▫ Different traffic sensitivity requirements
Any Medium Access Control protocol should aim at solving the problem of Coexisted CRNs enabling them to live together maintaining QoS senstivity and ensuring maximum resource utilization.
10Design Goals
MAC Protoco
l for CCRN
Increasing Spectrum Utilization
Weighted Load
Distribution among
Channels
QoS Awareness
Increasing System
Throughput
Decreasing Channel
Switching Rate
InterferenceMinimization
11Challenges
Development of a strong distribution mechanism, for the independent channel selection of heterogeneous CRNs
Limiting the effects of the non-cooperative mechanism to achieve fairness
Identification and maintenance of QoS sensitivity of numerous applications
Development of adaptive and dynamic medium access mechanism
Maintenance of effective channel usage through knowledgeable decision
12
State-of-the-art Works
13Related Work• Credit-Token based channel selection [B. Gao, Y. Yang, and J.-M. Park, A
credit-token-based spectrum etiquette framework for coexistence of heterogeneous cognitive radio networks Wireless Communications, INFOCOM, 2014 Proceedings IEEE]
▫ Enable spectrum sharing among distributed heterogeneous CR networks with equal priority
▫ Propose a centralized algorithm, not very feasible in heterogeneous CRN environment.
▫ The medium access and auction policy is not adaptable to network’s traffic load and traffic’s QoS requirements.
14Related Work• SHARE [Kaigui Bian, Jung-Min , Xiaojiang Du, and Xiaoming Li, and Sung Won Kim. Ecology-inspired
coexistence of heterogeneous wireless networks. Global Communications Conference (GLOBECOM), 2013 IEEE]
▫ Symbiotic Heterogeneous coexistence ARchitecturE (SHARE)
▫ Enable collaborative coexistence of heterogeneous CR networks over TV white space.
▫ Adopts the symbiotic relationships between heterogeneous organisms in a stable ecosystem.
▫ Continuous communication imposes higher protocol operation overhead
▫ Unstable system with no historical prediction▫ QoS awareness and SU activity on PU arrival are avoided.
15
•FMAC[Yanxiao Zhao; Min Song; ChunSheng Xin, "FMAC: A fair MAC protocol for coexisting cognitive radio networks," INFOCOM, 2013 Proceedings IEEE , December 2013]
▫ Pioneer on considering coexistence property. ▫ Distributed fair MAC for heterogeneous coexistence.
▫ channel allocation is done without usage pattern prediction, SU selects a channel based on only the current sensing result
▫ overall fairness in channel usage is not guaranteed.▫ QoS awareness is also avoided▫ The option of channel switching is totally disregarded.
Related Work
16Related Work
A SU1, SU2, SU3 wants to send packets
SU2 senses two channel as free, randomly selects
channel 1
SU2 sends one packet
SU2 keeps sensing and starve
SU1, SU3 senses two
channel as free, randomly
selects channel 3
Although SU3 has more critical data, it can randomly select high back-off value than SU2
PU packetsSU4 packets
• Starvation• Non judicious, random channel selection• No QoS awareness• No usage fairness• Low throughput• More interference
17Related WorkCredit Token
SHARE FMAC nQ WF-MAC
random WF-MAC
WF-MAC
Distributed x x
Dynamic x x x (partly) (partly) Channel Selection (partly) (partly) x x
Weighted Fairness x (partly) (partly) (partly) (partly)
Three-state sensing
x x
Learning x x x QoS Awareness
x x x x
18
Weighted Fair Medium Access Control (WF-MAC)
19ContributionA Distributed Quality of Service Aware Medium Access in Coexisting Cognitive Radio Networks
Multilevel weighted fair resource utilization is maintained by the SUs through:
• Knowledgeable channel selection, and• QoS aware channel access
Channel selection stability is achieved through two dimensional learning:
• Channel availability prediction, and• Channel utility perception
Provides a rational compensation between channel sharing and channel switching
20System Architecture• Infrastructure based Coexisting Cognitive Radio Network• Multiple CRNs• Two types of users
▫ Primary User (Licensed Users) ▫ Secondary Users (Unlicensed Users)
• | | number of licensed channels▫ Each channel is conditionally and opportunistically accessible by
the SUs
• Each SU is equipped with two radios▫ one for spectrum sensing▫ others is for data transmission
• Each CRN has one Base Station (BS)▫ Each SU of a specific CRN, share their sensing information
(cooperative sensing) with their BS after a specific time intervals.
21System Architecture
Figure: Coexisting Cognitive Radio Network
22Three State Sensing Model
• A three-state sensing model is used, where each busy state is further divided into state 1 (accessed by PU) and state 2 (accessed by SU), using a distance based estimation technique.
ss : signal that an SU transmitssp : signal that the PU transmits,Si : is the signal that an SU received ni : is the zero-mean additive white Gaussian noise (AWGN).
[Y. Zhao, M. Song, C. Xin, and M. Wadhwa, “Spectrum sensing based on three-state model to accomplish all-level fairness for co-existing multiple cognitive radio networks.” in IEEE INFOCOM, 2012]
23Control Packets• SUs exchange control messages over a common control channel (CCC).• There are different approaches for transmitting control messages for CRNs and also for CCC selection, like [12-15].
Figure : Different Types of Packets Transmitted over CCC
24
Table: QoS Aware Traffic Prioritization
[Wi-Fi Alliance. Wi-fi certified for wmm – support for multimedia applications with quality of service in wi-fi networks. Technical report, Wi-Fi Alliance, 2004]
Traffic Prioritization
25
Not
atio
n Ta
ble
26W
F-M
AC D
esig
n Co
mpo
nent
s
27
Weighted Fair Channel Selection
28Weighted Fair Channel Selection
• In coexisting CRN environment▫ each CRN functions in a distributive and non-cooperative
way.
• Every CRN should work towards maximizing the channel utilization▫ which can be achieved only by maximizing individual SUs’
utilization over the course of time.
• With legitimate knowledge of system’s current state will allow SUs to gain the finest and most rational channel distribution and maximize spectrum utilization.
29
Weighted Fair Channel Selection
• We are using two dimensional learning mechanism for channel selection▫ Perception based learning mechanism
▫ Channel Utility Perception Vector ▫ Arrival probability prediction
▫ Channel availability vector
30Weighted Fair Channel Selection
• Whenever an SU has some data to transmit, it sends a RCIV packet to its BS.
• Then the BS prepares channel information vector
Hi: {0, 1, 2}; Status of channel i : PU arrival rate over channel i : SU arrival rate over channel i : Perception utility of channel i
31
Weighted Fair Channel Selection
•Using the directives of and , the SU selects a channel from the channel set for contention based channel access.
: Perception utility of channel i : Probability of channel i being free : Probability that PU will not appear over Channel i : Probability that SU will not appear over Channel i : Probability thresholds
32Weighted Fair Channel Selection
33Channel Availability Prediction• The arrival pattern of PU and SU follows possion
distribution• Probability that no SU or PU will appear over the data
transmission time can be calculated▫ Also the overall probability of a channel being idle.The expected time
needed to transfer current packet of the SU over channel i
The PU arrival rate over channel iThe SU arrival rate over channel iProbability of channel being free
Channel Availability Vector
34
• The expected time an SU needs to transmit its current data packets derived using ▫ maximum achievable data rate βi of each channel ▫ average medium access delay in between two consecutive
data packet transmission,
Channel Availability Prediction
: The length of a single data packet : The number of packet SU wants to send : The expected value of back-off counter : The propagation delay : The length of single time slot
35
• The BS calculates the maximum achievable data rate βi of each channel▫ which is strongly related to the signal-to-noise-ratio▫ calculated using Shannon’s theorem
Channel Availability Prediction
Bi: Bandwidth of channel iSINRi : Signal to Interference-plus-Noise-Ratio on SU-BS transmission link over channel iSn, Sk: Received signal power from SU n, kNp: The noise powerd(n, r), d(k, r): Euclidean distance between the BS and SUs
36
Channel Availability Prediction• We are using a Auto Regression (AR) model of order Δ
to predict the arrival rates of each type of users.The PU arrival rate
The SU arrival rate
The autoregressive coefficient
The prediction error
[R. A. Yaffee and M. McGee, Introduction to Time Series Analysis and Forecasting: With Applications of SAS and SPSS, 1st ed. Orlando, FL, USA: Academic Press, Inc., 2000][Z. Wen, T. Luo, W. Xiang, S. Majhi, and Y. Ma, “Autoregressive spectrum hole prediction model for cognitive radio systems,” in Communications Workshops, 2008. ICC Workshops ’08. IEEE International Conference on, Beijing, May 2008, pp. 154–157]
37
Channel Utility Perception
• We have adopted a perception based learning model which helps the BSs of different CRNs ▫ to build perception about each spectrum bands
by observing the utility gain and payoffs experienced by SUs over the course of time.
• Each SU only updates its selected channel’s utility• BS will aggregate these utilities experienced by
several SUs over different channels.
38
Channel Utility Perception•The set of possible outcomes experienced by
SU over a selected channel is defined as:
39
•uc is the utility perception of channel c, which is updated after each usage outcomes
Channel Utility Perception
: The constant gain received on successful transmission : The constant payoff on collision with SU : The constant payoff for channel switching
40
• Every BS initially sets each entry of a small constant value which implies no biasness towards any particular channel.
• After receiving perception utilities form SUs, respective channels’ perception utilities is updated by the BS.
Channel Utility Perception
Weighting factor
Number of SUs are accessing same channel i
41Weighted Fair Channel Selection
42
QoS Aware Medium Access Control
43QoS Aware Medium Access
44
QoS Aware Medium Access• QoS-aware Contention Window CWρ selection policy
is shown below
• The back-off counter bρ is selected by taking a random number between [1, CWρ].
: Number of times the current data packet is retransmitted because collision with SU or any kind of bit-error. : The number of times SU was penalized by PUs
45Q
oS A
war
e M
ediu
m
Acce
ss
46
Channel Switching Mechanism
47
Channel Switching Mechanism
• Secondary users schedule their spectrum usage in order to maximize spectrum utilization or throughput.
• To do that, the SUs should have the ability to rationalize between channel access or switching in a smart way.
48Rationalization of Channel Switching
• The expected throughput gain of the SU on the time of channel selection
• Several retransmissions or reappearance will scale down the expected throughput
• The SU should maintain θC ≥ θth ▫θth is the throughput threshold
: The transmission time over channel C for η number of packets : The probability being idle : Achievable bandwidth of channel c
κ: Constant, value of which is application dependentρ: Traffic priority
49Rationalization of Channel Switching• The SU will request for CIV from it’s BS and will select
new channel for transmission• But it has to define new channel set considering
channel switching cost
The channel switching cost
50
Performance Evaluation
51Simulation Environment
• We used network simulator version -3 (NS-3)[16][17] as simulation tool and conduct several experiments for the performance analysis of our proposed protocol, WF-MAC.
• Compared its performances with • FMAC[35]
• Two versions of our work: • random WF-MAC (with random selection of
channels)• nQ WF-MAC (that avoids QoS awareness)
52
Sim
ulat
ion
Para
met
ers
53Random Scenario
54
• We will use following metrics for evaluating our performance:
▫Throughput for SUs, Calculates the average number of data bits the SUs transmit per
seconds to their BSs over their active periods
▫Average medium access delay, Defines the average time taken for a secondary user to get
access of the medium before transmitting a packet
▫Protocol operation overhead, Measures the amount of control bytes exchanged per successful
data byte transmission, the portion of cost a MAC protocol pays for each byte of data transmission
▫Channel selection percentage Measures the average percentage of selection from each
category of channel (low, mid, high) over the total simulation period.
Performance Metrics
55
• We will use following metrics for evaluating our performance:▫ Integrated performance improvement
Measures the integrated performance of the studied protocols as follows,
which quantifies the cost compensation for the increased throughput and reduced medium access delay performances.
▫Medium access delay of traffic classes Calculate the average medium access delay of each type
of traffic class over the active periods of the SUs.
Performance Metrics
56
Impact of Increasing Number of CRNs
57
Impact of increasing number of CRNs
Simulation Result
WF-MAC performs• 48.98% over the FMAC• 18.73% over random WF-MAC• 17.64% nQ WF-MAC protocol
WF-MAC experiences, on an average, 43.33% less delay than FMAC
58
Impact of increasing number of CRNs
Simulation Result
Additional load of RCIV (20
bytes) and CIV (260 bytes)
notably increase (42% -43.2%)
the protocol operation overhead
of WF-MAC
With the increasing CRN, FMAC
decreases its integrated
performance greatly, 72.13%
than WF-MAC
59
Impact of increasing number of CRNs
Simulation ResultWeighted
Fair Channel Selection
QoS Aware Medium Access
60
Impact of Increasing Number of PUs
61
Impact of increasing number of PUs
Simulation Result
FMAC earns 26.84%
less throughput
than WF-MAC
WF-MAC obtains SU
throughput
• 41.65% than nQ WF-MAC,
• 68.65% than random WF-
MAC,• 88.56% than FAMC
The gap between the performance of WF-MAC and FMAC increases from 33%to 64%
62
Impact of increasing number of PUs
Simulation Result
WF-MAC has huge
performance improvements
over FMAC and random WF-MAC
Almost double (48.49%)
63
Impact of increasing number of PUs
Simulation Result
QoS Aware Medium Access
64
Impact of Increasing Number of SUs
65
Impact of increasing number of SUs
Simulation Result
WF-MAC gain higher throughput • 34% than FMAC, • 10% random WF-MAC ,• 16% nQ WF-MAC
66
Impact of increasing number of SUs
Simulation Result
67
Impact of increasing number of SUs
Simulation Result
Weighted Fair
Channel Selection
QoS Aware Medium Access
68References[1] D. Goldman, “FCC scrambles to cope with data
avalanche,” http://money.cnn.com/2011/12/29/technology/whitespace spectrum/index.html, accessed on May accessed on May 2015
[2] Facilitating Opportunities for Flexible, Efficient, and Reliable Spectrum Use Employing Cognitive Radio Technologies, Federal Communications Commission (FCC), Washington, D.C. 20554, December 2003
[3] G. M. Peter Steenkiste, Douglas Sicker and D. Raychaudhuri, “Future directions in cognitive radio network research,” NSF Workshop, Tech. Rep., 2009
[4] C. R. W. Group, Quantifying the Benefits of Cognitive Radio, The Software Defined Radio Forum Inc., 2010, WINNF-09-P-0012-V1.0.0.
[5] W. I. Forum, “Defining cognitive radio (CR) and dynamic spectrum access (DSA),” http://www.wirelessinnovation.org/defining cr and dsa, accessed on April 2015.
[6] S. Pollin, M. Timmers, and L. Van der Perre, “Emerging standards for smart radios: Enabling tomorrow’s operation,” in Software Defined Radios, 1st ed., ser. Signals and Communication Technology. Springer Netherlands, 2011, pp. 11–35
[7] S. D. R. F. Inc., “SDR: Software defined radio,” https://forums.hak5.org/index. php?/forum/81-sdr-software-defined-radio , accessed on March 2015
[8] B. Gao, Y. Yang, and J.-M. Park, A credit-token-based spectrum etiquette framework for coexistence of heterogeneous cognitive radio networks Wireless
Communications, INFOCOM, 2014 Proceedings IEEE
[9] Kaigui Bian, Jung-Min , Xiaojiang Du, and Xiaoming Li, and Sung Won Kim. Ecology-inspired coexistence of heterogeneous wireless networks. Global Communications Conference (GLOBECOM), 2013 IEEE
[10] Yanxiao Zhao; Min Song; ChunSheng Xin, "FMAC: A fair MAC protocol for coexisting cognitive radio networks," INFOCOM, 2013 Proceedings IEEE , December 2013
[11] Y. Zhao, M. Song, C. Xin, and M. Wadhwa, “Spectrum sensing based on three-state model to accomplish all-level fairness for co-existing multiple cognitive radio networks.” in IEEE INFOCOM, 2012
[12] L. Lazos, S. Liu, and M. Krunz, “Spectrum opportunity-based control channel assignment in cognitive radio networks,” in Proceedings of the 6th Annual IEEE Communications Society Conference on Sensor, Mesh and Ad Hoc Communications and Networks, ser. SECON’09. Piscataway, NJ, USA: IEEE Press, 2009, pp. 135–143.
[13] K. Chowdhury and I. Akyldiz, “Ofdm-based common control channel design for cognitive radio ad hoc networks,” Mobile Computing, IEEE Transactions on, vol. 10, no. 2, pp. 228–238, Feb 2011.
69References[14] M. S. Miazi, M. Tabassum, M. Razzaque,
and M. Abdullah-Al-Wadud, “An energyefficient common control channel selection mechanism for cognitive radio ad hoc networks,” Annals of telecommunications, vol. 70, no. 1-2, pp. 11–28, 2015.
[15] Y. Zhang, G. Yu, Q. Li, H. Wang, X. Zhu, and B. Wang, “Channel-hopping-based communication rendezvous in cognitive radio networks,” Networking, IEEE/ACM Transactions on, vol. 22, no. 3, pp. 889–902, June 2014.
[16] “Network simulator-3,” https://www.nsnam.org/, accessed on: January 2015.
[17] N. Kamoltham, K. Nakorn, and K. Rojviboonchai, “From NS-2 to NS-3 : Implementation and evaluation,” in Computing Communications and Applications Conference (ComComAp), Hong Kong, January 2012, pp. 35–40.
[18] R. A. Yaffee and M. McGee, Introduction to Time Series Analysis and Forecasting: With Applications of SAS and SPSS, 1st ed. Orlando, FL, USA: Academic Press, Inc., 2000]
[19] Z. Wen, T. Luo, W. Xiang, S. Majhi, and Y. Ma, “Autoregressive spectrum hole prediction model for cognitive radio systems,” in Communications Workshops, 2008. ICC Workshops ’08. IEEE International Conference on, Beijing, May 2008, pp. 154–157]
70
List Of Publications
[1] ——————–, “QoS Aware Weighted-Fair Medium Access Control Protocol for Coexisting Cognitive Radio Networks,” Submitted to EURASIP Journal on Wireless Communications and Networking, April 2015
[2] ——————–, “Traffic Priority and Load Adaptive MAC Protocol for Body Sensor Network with QoS Provisioning,” International Journal of Distributed Sensor Networks on the issue of “Recent Advances in Energy-Efficient Sensor Networks (EESN)”, Article ID 205192, 9 pages, vol.2013, February, 2013 (doi:10.1155/2013/205192)
71
Conclusion• A weighted fair opportunistic medium access control protocol, WF-MAC, has been developed for QoS aware traffic delivery in coexisting cognitive radio networks.
• Fully distributed and driven by traffic class priorities and opportunistic spectrum qualities.
• The two dimensional learning mechanism consisting of perception learning and channel availability prediction helps our WF-MAC to achieve as high as 88.56% and 64% improvements in throughput and medium access delay, respectively, compared to FMAC for varying arrival rates of primary users.
Thank You
Any Questions ?
73
Channel Selection Stability
74
Channel Selection Stability
Simulation Result
75
Channel Selection Stability
Simulation Result
76
Three-State Sensing Model
Figure: Distance estimation of three-state sensing model
Three-state sensing model uses two stage decision policy: • In the first stage, energy detection methodology identify whether a channel is idle or busy. • The received signal is then further analyzed based on a distance based estimation technique, aiming to effectively differentiate PUs signals from SUs, using the statistical model of the locations of SUs and known locations of PUs.
77Exponentially Weighted Moving Average• More recent returns have greater weight on the
variance.
• Relatively little data needs to be stored
• An exponentially weighted moving average can be defined on any time series of data.
• The simplest form of exponential smoothing is given by the formula:
yt = α yt + (1 − α) y(t − 1)
where α is the smoothing factor, and 0 < α < 1.
78Auto-Regression (AR) Model• An AR model expresses a time series as a linear
function of its past values. • The order of the AR model tells how many lagged past
values are included. • The simplest AR model is the first-order autoregressive,
AR(1),
yt = a1yt-1 + εt
where, yt is the mean-adjusted series in time t, yt-1 is the series in the previous interval, {|at| < 1} is the lag-1 autoregressive coefficient, and εt is the noise.
79
AR vs EWMA
80
Recommended