Compressive Spectrum Sensing Based on Spectral Shape Feature Detection

In this paper, we address sparsity-based spectrum sensing for Cognitive Radio (CR) applications. Motivated by the sparsity described by the low spectral occupancy of the licensed radios, the proposed approach utilizes the novel Compressive Sensing (CS) technique to alleviate the sampling burden in CR when processing very wide bandwidth. Instead of detecting underutilized subbands of the radio spectrum, this paper propose a feature-based strategy to detect the licensed holder activity from compressive measurements. The procedure follows the framework of correlation matching, changing the traditional single frequency scan to a spectral scan with the a priori known spectral shape of the licensed holder. In addition to the frequencylocation estimate, the proposed technique is able to provide a power-level estimate and an estimation of the angle-of-arrival (AoA) of the primary users by circumventing the complex nonlinear CS reconstruction.