Sparse signal detection with spatial diversity using multi-rate sampling

We propose a spectrum sensing system consisting of K spatially distributed cognitive radios linked to a fusion centre. Each cognitive radio samples a wideband sparse multi-band signal, where each band comes from a distant device, at a random rate below Nyquist. Estimations of the signal support are done locally on undersampled data, then are processed at the data fusion center which finally outputs the support of the original signal. The proposed system is resilient and offers substantial gains in terms of detection by exploiting spatial diversity. The overall sampling rate remains moderate and the local sampling rates are far below that of Nyquist, relaxing the sampling rate constraints on ADCs, a traditional limiting factor in wideband sampling.