Collaborative compressed spectrum sensing: what if spectrum is not sparse?

The problem of detecting the failure of compressed spectrum sensing in wideband cognitive radio is considered. When the primary users occupy most sub-channels, the compressed spectrum reconstruction will fail owing to the non-sparsity of the spectrum. A Gaussian process (GP) framework is proposed to model the compressed spectrum reconstruction and exploit the correlation of model parameters to detect the failure of compressed reconstruction owing to non-sparse spectrum. The simulation results demonstrate that the proposed collaborative non-sparsity protection (CNSP) scheme can effectively identify the failure of spectrum sensing and avoid interference to primary users.