Distributed Wideband Sensing for Faded Dynamic Spectrum Access with Changing Occupancy

We propose a distributed compressive sampling technique for cooperative wideband spectrum sensing that requires lesser numbers of measurements while overcoming time-variability of spectrum occupancy and the hidden terminal problem. First, we prove that the wideband spectrum occupancy information can almost surely be recovered with a reduced number of spectrum measurements. Second, we propose non-uniform sensing matrix design that exploits the heterogeneity in the wideband spectrum access to further improve the spectrum sensing recovery accuracy. Using simulations, we confirm our theoretic results and show that cooperation leads to high detection probability, even with each secondary user taking only a small number of measurements. We also show that it is sufficient to consider a subset of close-by secondary users to obtain comparable performances.