Uniform Linear Array Based Spectrum Sensing from Sub-Nyquist Samples

With the emergence of Cognitive Radios (CRs), that aim at solving the spectrum scarcity issue, the traditional task of spectrum sensing has recently been revisited. Blind sub-Nyquist sampling and reconstruction methods of multiband signals have been proposed, alleviating the burden of both the analog and digital sides. In this work, we propose a new sub-Nyquist sampling system composed of sensors lying in a uniform linear array (ULA) that adopts some of the concepts of the Modulated Wideband Converter (MWC). Our system overcomes two practical issues of the MWC: the challenging choice of mixing functions which intentionally aliases the signal, and the introduction of the same sensor noise to all of the system channels. We provide two carrier frequencies recovery algorithms, and show how the signal can be reconstructed once these are estimated. We derive bounds for the minimal number of sensors and minimal sampling rate required for perfect signal reconstruction. Simulations show that for an equal number of channels or sensors and an identical sampling rate, our system outperforms the traditional MWC in terms of reconstruction performance.