Low-Complexity Census-Based Collaborative Compressed Spectrum Sensing for Cognitive D2D Communications

The compressed spectrum sensing problem for cognitive radio (CR) spectrum sharing in device-to-device (D2D) communications is investigated, with an emphasis on collaborative sensing strategies for multi-user CR-based D2D networks. Because the D2D users are assumed to be geographically close to each other, the same spectral occupancy is expected, which can be exploited in the sensing algorithm design. We first investigate a single user compressed spectrum sensing algorithm, where the successive fast iterative shrinkage-thresholding algorithm (FISTA) is employed. The successive FISTA-based single-user sensing algorithm is then used to develop two collaborative compressed spectrum sensing schemes, namely Equal-Gain Combining (EGC) and Census-Weighted Detection Results Combining (CWDRC). It is demonstrated that both algorithms are effective in low signal-to-noise ratio (SNR) conditions. We further show that the census-based CWDRC algorithm significantly reduces system overhead compared to other collaborative sensing strategies with only a small degradation in the probability of detection. Thus, CWDRC is attractive for collaborative spectrum sensing in CR-based D2D systems.