Channel Energy Statistics Modeling and Threshold Adaption in Compressive Spectrum Sensing

Abstract

Compressive spectrum sensing (CSS) techniques alleviate the demand of high-speed sampling in wideband spectrum sensing for cognitive radio systems. Known existing literature discusses threshold adaption schemes to achieve optimal performance of channel occupancy detection in conventional non-compressive spectrum sensing scenario. However, in the CSS case, it is found that the channel energy statistics and optimal threshold not only depend on noise energy in channel but also compression ratio, the selection of recovery algorithms, etc. Therefore, we postulate a statistical model of channel energy in CSS and propose a practical threshold adaption scheme aiming to achieve constant target false alarm rate. The validity of the postulated channel energy model is verified by learning the parameters of a Mixture Model and aligning with empirical distributions. Finally, performance of the proposed threshold adaption scheme is presented and discussed.