An Efficient Non-sparsity Protection Scheme for Collaborative Compressed Spectrum Sensing.

The problems of evaluating the performance of local compressed reconstruction and detecting the failure of compressed spectrum sensing are considered in this paper. When the monitored spectrum is not sparse, spectrum sensing methods based on compressed sensing will fail to obtain correct occupancy status, and then the operations of unlicensed spectrum access will cause severe interference to licensed users. To alleviate this problem, in this paper, a collaborative non-sparsity protection scheme based on decision correlation is proposed, in which the correlation between local decisions of nearby sensing nodes is exploited to evaluate the reliability of local decision and identify the failure of compressed spectrum sensing due to the non-sparsity of the spectrum. Furthermore, the reason of adopting correlation metric is elucidated and a simulation-based approach is presented to determine the proper threshold for correlation metric, which offers useful guideline for practical application. Numerical simulation results demonstrate that the proposed scheme can effectively detect the failure of compressed spectrum sensing and avoid severe interference to licensed users. © 2013 ACADEMY PUBLISHER.