Collaborative Compressive Spectrum Sensing with Missing Observations for Cognitive Radio Networks Conference Paper uri icon


  • © 2014 IEEE. Spectrum sensing, which seeks to detect the unoccupied channels or spectrum holes, is the first task for ensuring the functionality of Cognitive Radio (CR) system. But considering hardware limitation, each CR node can only obtain limited information about the spectrum usage on whole spectrum channel. Thus, by exploiting compressive sensing (CS) technology, collaborative compressive sensing for spectrum sensing was proposed to solve this problem. However, due to channel fading and transmission power limitation, fusion center (FC) usually cannot receive complete measurements from all CR nodes. Thus, now the problem is how to obtain the complete information of spectrum usage from the incomplete measurements. Although previous matrix completion recovery (MCR) algorithm has focused on this issue, it's less applicable in the noise environment and the situation of large number of measurements (observations) are missing. Thus, we propose a new recovery algorithm for spectrum sensing with missing observations in this paper. Different to MCR algorithm, our proposed algorithm needs not to run matrix completion algorithms but can detect the occupied channels from the incomplete observations directly. Moreover, our method outperforms MCR algorithm in both the situations of noise corruption and large number of observations are missing. Furthermore, we proposed a sparsity adaptive based dynamic compressive spectrum sensing algorithm which is aiming at solving the problem of spectrum sensing in the dynamic environment. This dynamic algorithm focus on recovering the recent changes which are the newly occupied channels or the released channels. Compared to previous dynamic compressive spectrum sensing (DCSS) algorithm which can only detect single channel change once a time, our method which can detect multiple channels changes is more suitable for the application. Simulation results will also validate the effectiveness of all our proposed schemes.

author list (cited authors)

  • Jin, S., & Zhang, X. i.

citation count

  • 0

publication date

  • December 2014