Data-selective cooperative spectrum sensing based on imperfect information exchange

This paper examines the use of set-membership adaptive filters (SMAF) to distributed detection for spectrum sensing when only imperfect estimates of data statistics are available. The first and second central moments of the energy at each sensor node are calculated as sample averages, and the impact of such estimates on the detection performance is evaluated. The use of SMAF for coefficient update allows significant savings in computational complexity per node and also provides good indication of when sharing data is beneficial. Such features are exploited to propose a cooperative strategy of selected energy estimates. The paper also evaluates the impact of packet loss in the control channel on detection performance. For the simulated scenarios, data selection based on the set-membership approach proved to be a valuable tool for distributed detection in cognitive radio applications due to enhanced robustness, power savings, and increased probability of detection.