Spectrum Sensing in Emergency Cognitive Radio Ad Hoc Networks (CRAHNs) : A Multi-Layer Approach
Sasirekha GVK, ,Supervisor: Prof. Jyotsna Bapat, IIIT Bangalore
Requirements of Emergency CRAHNs:
•Accuracy
•Resource efficiency
•Low latency in the delivery of packets,
•Adaptive to varying number of SUs,
•Adaptive to varying SNR conditions,
•Uniform battery consumption
•Resilience to Byzantine attacks
SNR
Threshold
Sensing Mechanism
Local decisions, accuracy
,
Fusion Rule
NumberOf
SensingSUs
Sensing timeFrequency
of sensing
PHY LINK
Global decisions, accuracy
,
Performance
Literature surveyCollaborative spectrum sensing
1. Amir Ghasemi and Elvino S. Sousa,
2. Wei Zhang, Rajan K. Mallik, Khaled Ben Letaief
3.Clancy
4. L. Chen, J. Wang, S. Li,
5. Yunfei Chen
Static/Reactive methods using ‘OR’ based fusion, Civilian Networks
Considering only some parameters for optimization
Cognitive Radio Ad hoc Networks
Ian F. Akyildiz, Won-Yeol Lee, Kaushik R. Chowdhury, Protocol stack, routing, transport and high level architecture
Emergency NetworksAdaptive Ad-hoc Free Band Wireless Communications Requirements in
general
IEEE Standards IEEE 802.22 (Shell Hammer) Regional Area Networks in TV band
Our proposal proactive, dynamic, LRT based (better immunity against Byzantine attacks) meeting sensing requirements for emergency networks
Multi-Layer Framework
Focus of the research
Confidence
Link Layer
Blind/Semi-blindSpectrum Sensing
Averaging AndFinal
DecisionLogic
Decision
Rx_Signal
Threshold
Data Fusionwith opt. KEstimator
Soft/Hard Decision from other users
Cognitive Radio Receiver
Front End
Physical Layer
Adaptive Thresholding
Group Decision
Sensing Scheduler
Being a Multi-Layer Multi-Parameter optimization problem tackled as 2 levels•Level 1: Local Optimization: Spectrum sensing method, time, frequency•Level 2: Global Optimization: Data Fusion, Optimal number of Sensing CRs•Cross Layer: Adaptation of local sensing threshold based on Global Decisions
Results• Estimation of smallest number of sensing CRs for a targeted accuracy.
• Algorithm for adapting the number of sensing SUs in changing environments; i.e. network size and SNR. Proposed for centralized and distributed spectrum sensing.
• Algorithm for adapting threshold for local energy detection based on global group decisions.
• Application of evolutionary game theory for behavioral modeling of the network.
0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 10.85
0.9
0.95
1
1.05
Qd desired
Qd
actu
al
Qd actual versus Qd desired for various sensitivites
reference
-3%+3%
(Pd,Pf)=0.4,0.1
(Pd,Pf)=0.5,0.15
(Pd,Pf)=0.6,0.25(Pd,Pf)=0.76,0.4
(Pd,Pf)=0.85,0.5 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 104
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
iterations
Variance of energy spent,Payoff Qd, probability of sense of an SU with Qd target=0.9, Group SNR 0dB, Event 2 of Table I at iterations=10000
Norm
aliz
ed v
alu
e o
f variance /
pro
babili
ty
Normalized variance of energy spent across SUs
Probability of detect of fused dataProbability of sense of an SU
Sample Results on the Estimation of minimal no. of CRS and Adaptation of CRs
Open Issues
Cognitive Radio Ad hoc Network
Time synchronization
Optimized Link State Routing
Co-operative Spectrum Sensing
Com
mon C
ontrol Channel
Spectrum Allocation
Security •Provision of Common Control Channel
•Integration of all the layers
•Security Related Issues•Byzantine attacks•Primary User Emulation Attacks•Trustworthiness/ Authentication
SU
SU
SUSU
Coordinator
Centralized Architecture
SU
SU
SU
SU
SU
Distributed Architecture
Cognitive Radios : Secondary Users (SUs)Dynamic Spectrum Access
•Spectrum Sensing Local & Collaborative •Spectrum Allocation•Spectrum Mobility
Application Scenarios
PU
[f1 f2][f3 f4 f5 f6]
[fr-2 fr-1]
[fr]
Mobile CRAHNScenario model
PU PU
PU
•Military Networks•Disaster Management
Features:• Nomadic Mobility• Group Signal to Noise Ratio• Collaborative Spectrum Sensing
PHY LINK Performance Metrics
SNR
Threshold
Sensing Mechanism
ChannelModel
Local decisions,
Pdi
, Pfi
Fusion Rule
NumberOf
SensingSUs
Risk
From ith SU
From other (K-1) SUs
PUUsage pattern
Level 1 OptimizationLevel 2 Optimization
Sensing time
Frequency of sensing
Qdk
Qfk
Ik
k F fk D dkR C Q C Q C
k kI 1 R
k k k
k
J αI 1 α η
N k0 α 1,η
N
Two levels of optimization
Confidence
)λ(Yβ-ttt
tte1
1λYfz
t
2t
t1t λ
eEμλλ
)z1(zeμ2λλ tttt1t
Adaptive Threshold
Adaptive Threshold based on Group Decisions
)P,P,k(fQ~
f
~
dd
QQ kminK desired_dd
0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 10.85
0.9
0.95
1
1.05
Qd desired
Qd
actu
al
Qd actual versus Qd desired for various sensitivites
reference
-3%+3%
(Pd,Pf)=0.4,0.1
(Pd,Pf)=0.5,0.15
(Pd,Pf)=0.6,0.25(Pd,Pf)=0.76,0.4
(Pd,Pf)=0.85,0.5
Group SNR-> Pd_av, Pf_av-> K
Estimation of optimal number of CRs required for sensing for targeted accuracy
Behavioral ModelInteraction between autonomous CRs modeled
using game theory
PoliciesFrequencies to sense
Who should be the coordinator? Authenticate the entry into network
Implementation (Protocols)Adaptive System Design
Levels Of Abstraction
Ref: http: //www.ir.bbn.com/~ramanath/pdf/rfc-vision.pdf
Approaches of Analysis (Our Contributions)• Iterative Game (pot luck party) ---- Penalty• Evolutionary Game based on Replicator Dynamics --- Reward• Public Good Game ---Reward
• How many should sense? ---- K• Who should sense?• Assuming proactive spectrum sensing in the period quiet period
Game theoretical modeling
Adaptive Proactive Implementation Model: Centralized Architecture
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 104
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
iterations
Variance of energy spent,Payoff Qd, probability of sense of an SU with Qd target=0.9, Group SNR 0dB, Event 2 of Table I at iterations=10000
Nor
mal
ized
val
ue o
f var
ianc
e / p
roba
bilit
y
Normalized variance of energy spent across SUs
Probability of detect of fused dataProbability of sense of an SU
s avP _Ps _ av s _ avJ α I 1 α 1 P
Utility Function
Decentralized Architecture
)k(J)k(MaxK
1 constant a is ε where
εJ)k(MinK '
J (1 ) I
1J C Q C Q
2 ND d F f
0 10 20 30 40 50 60 70 8010
0
101
102
103
104
105
N
No. o
f Multi
plicatio
ns
Computational Complexity Vs. N
Classical Iterative Algorithm
Proposed Algorithm
1. Sasirekha GVK, Jyotsna Bapat, “ Adaptive Model based on Proactive Spectrum Sensing for Emergency Cognitive Ad hoc Networks”, CROWNCOM 2012, Stockholm, Sweden
2. Sasirekha GVK, Jyotsna Bapat , “Optimal Number of Sensors in Energy Efficient Distributed Spectrum Sensing”, CogART 2010. 3rd International Workshop on Cognitive Radio and Advanced Spectrum Management. In conjunction with ISABEL 2010. November 08-10, 2010, ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5702906
3. Sasirekha GVK, Jyotsna Bapat, “Optimal Spectrum Sensing in Cognitive Adhoc Networks: A Multi-Layer Frame Work”,
CogART 2011 Proceedings of the 4th International Conference on Cognitive Radio and Advanced Spectrum Management
Article No. 31, ACM, ISBN: 978-1-4503-0912-7 doi>10.1145/2093256.20932874. Sasirekha GVK and Jyotsna Bapat, “Evolutionary Game Theory based Collaborative Sensing Model in Emergency
CRAHNs," Journal of Electrical and Computer Engineering, Hindawi Publishing Corporation, Special issue "Advances in Cognitive Radio Ad Hoc Networks“, (accepted)
5. Sasirekha GVK ,George Mathew Tharakan, Jyotsna Bapat, “Energy Control Game Model for Dynamic Spectrum Scanning”, IJAACS, Inderscience, 2012, DOI: 10.1504/IJAACS.2012.046280
6. Sasirekha GVK, Jyotsna Bapat, “Cognitive Radios: A Technology for 4G Mobile Terminals”, Third Innovative Conference on Embedded Systems, Mobile Communication and Computing, 11th- 14th August, 2008, Infosys, Mysore, India, http://www.pes.edu/mcnc/icemc2/
7. Rajagopal Sreenivasan, Sasirekha GVK and Jyotsna Bapat, “Adaptive Threshold based on Group Decisions for Distributed Spectrum Sensing in Cognitive Adhoc Networks”, Wimone 2010 8. Rajagopal Sreenivasan, Sasirekha GVK and Jyotsna Bapat, “Adaptive Threshold based on Group intelligence”,
International Journal of Computer Networks and Communications , AIRCC,May 20119. Sasirekha GVK, Jyotsna Bapat IGI-CRN Book Chapter # 4: “Spectrum Sensing in Emergency Cognitive Radio Ad Hoc
Networks”, Cognitive Radio Technology Applications for Wireless and Mobile Ad hoc Networks. IGI Global (under (under review)review)
Papers Published on Research Topic