Autonomous compressive spectrum sensing approach for 3.5 GHz shared spectrum

The underutilized 3.5 GHz shared spectrum poses an excellent opportunity and potential for more intensive secondary usage by innovative applications/services. To find more spectral holes, a wide portion of spectrum must be sensed, which requires high sampling rates and a lot of measurements to be processed. Compressive sensing (CS) has recently become one of the promising techniques to deal with the sampling rate bottleneck of the wideband spectrum sensing. However, there are two significant challenges in the implementation of CS based wideband spectrum sensing: 1) no apriori knowledge of users activity statistics and 2) the varying bandwidth of channels and power levels. To address these issues, we proposed an autonomous compressive spectrum sensing approach that enables a secondary user to choose the number of measurements automatically, while the exact wideband signal reconstruction is guaranteed without assumption on spectral sparsity or channel characteristics. Specifically, the compressive measurements are collected block-by-block while the spectral is gradually reconstructed and the measurements collection process can be terminated once the variation of the Euclidean distance among the sequence of recovery solutions falls below a desired tolerance.