Spectrum Sensing and Opportunity Detection in Cognitive Radio Networks

1. Chapter 4 Spectrum Sensing and Identification “Cognitive Radio Communications and Networks: Principles and Practice” By A. M. Wyglinski, M. Nekovee, Y. T. Hou (Elsevier, December 2009)

2. Outline • Introduction • Primary Signal Detection • Spectrum Opportunities Detection • Performance vs. Constraint • Sensing Accuracy vs. Sensing Overhead

3. Introduction • Limited supply

4. Introduction • Growing demand

5. Current Policy & Spectrum Scarcity

6. Spectrum Opportunitiesin Space, Time, & Frequency (Credit: DARPA XG) (Credit: ACSP Cornell)

7. Primary Signal Detection • Choice of detectors • Criteria: • Bayesian • Neyman-Pearson • Parameter settings? • Energy detection • Pros: easily implemented; minimal assumptions • Cons: poor performance with noise uncertainty and with multiple secondary users Performance ∼ 1/SNR2 at low SNR

8. Choice of Detectors - Cyclic Detectors (2) • Exploit guard bands in frequency, known carriers, data rates, modulation type • Pros: • fc, Ts easy to detect via square-law devices, or cyclic approaches • Cyclic approaches useful when σ2n is unknown (avoid SNR wall) • Easily implemented via FFTs • Cons: • Timing and frequency jitter can be detrimental • Requires long integration times • RF non-linearities; Spectral leakage (ACI).

9. Choice of Detectors: Matched Filter (3) • Exploit pilots or sync (PN) sequences in primary (WRAN 802.22) • Pros: • Correlation detection is usually better than energy detection. • Performance ∼ 1/SNR at low SNR • Cons: • fading may null pilot; need to cope with time and freq sync

10. Other Detectors • Receiver leakage Wild-Ramachandran, Dyspan’05 • Signal correlation Zeng et al, PIMRC’07 • Fast fading Larson-Regnoli, CommLett’07 • Multiple antennas Pandhripande-Linnartz, ICC’07 • HMM classifier Kyouwoong et al, Dyspan’07 • Wavelet-based Tian-Giannakis, CrownCom’06 • Multi-resolution sensing Neihart-Roy-Allstot, ISCAS’07 • Compressed sensing Tian-Giannakis, ICASSP’07

11. Spectrum Opportunities Detection • A channel is an opportunity for A − B if • the transmission from A to B can succeed • the interference power to primary is below a prescribed level

12. Spectrum Opportunity: Definition • A channel is an opportunity for A − B if • the transmission from A to B can succeed • the interference power to primary is below a prescribed level

13. Spectrum Opportunity: Definition • A channel is an opportunity for A − B if • the transmission from A to B can succeed • the interference power to primary is below a prescribed level

14. Spectrum Opportunity: Properties • Determined by both transmitting and receiving activities of primary users. • Asymmetric (an opportunity for A−B may not be one for B−A).

15. Detection of Primary Receivers • rI: interference range, Rp: primary tx range, rD: detection range • Detecting primary Rx within rI by detecting primary Tx within rD

16. Detecting Primary Signals (LBT) • rD: detection range. • H0: no primary Tx within rD, H1: alternative. • False alarms and miss detections occur due to noise and fading.

17. From Detecting Signal to Detecting Opportunity • H0: opportunity, H1: alternative. • Even with perfect ears, exposed Tx(X) ⇒ FA, hidden Rx(Y) ⇒ MD. • Adjusting detection range rD leads to different operating points.

18. When Is Detecting Signal = Detecting Opportunity? • A Necessary and Sufficient Condition: • NS condition: ∀X ∈ Ptx(A) ∩ Pctx(B), its receivers are in Prx(A) • Perfect detection achieved when detecting Ptx(A) ∪ Ptx(B)

19. Miss Detection May not Lead to Collision • There is no primary receiver around A • There are primary transmitters around B

20. Miss Detection May Lead to Success • There are primary receivers around A • There is no primary transmitter around B

21. Correctly Identified Opportunity May Not Lead to Success • Successful data transmission and failed ACK

22. Performance vs. Constraint • Performance • Optimal under relaxed constraint on the average number of active arms. • Asymptotically optimal (N →∞ w. M/N fixed) under certain conditions. • Near optimal performance observed from extensive numerical examples.

23. Performance vs. Constraint • Two Models • Global Interference Model • Local Interference Model

24. Performance vs. Constraint Throughput comparison.

25. Sensing Accuracy vs. Sensing Overhead Optimal sensing time: efficiency η versus sensing window length n for various SNRs and PMD.

26. Sensing Accuracy vs. Sensing Overhead Optimal sensing time: efficiency η and optimal window length n∗/N versus slot length N.

27. Chapter 4 Summary The following topics have been covered: • Different types of detectors for primary signal detection • Detection of spectrum opportunities based on the detection of primary signals. • The trade-off between performance and interference constraint. • The trade-off between sensing accuracy and sensing overhead.