A Novel Sub-Nyquist Multiband Signal Detection Algorithm for Cognitive Radio

Abstract

Wideband spectrum sensing (WSS) is an essential technology for cognitive radio. However, the sampling rate is still a bottleneck of WSS. Several sub-Nyquist sensing methods have been proposed. These technologies deteriorate in the low signal to noise ratio (SNR) regime or suffer high computational complexity. In this paper, we propose a novel sub-Nyquist WSS method based on Multi-coset (MC) sampling. We design a simple SNR-robust and low-complexity multiband signal detection algorithm. In particular, the proposed method differs the commonly used detection algorithms which are based on energy detection (ED), matched filter (MF) or cyclostationary detection (CD). We exploit the linear recurrent relation between the locations of nonzero frequencies and the DFT of the arithmetic-shifted subsampled signals. These relations can be uniquely expressed by a series of the so-called frequency locator polynomials (FLPs). The scalar of the relations is related to the bandwidths of the subsignals. Basing on this, we propose a detector for sparse multiband signals along with the method estimating carrier frequency and bandwidth. The detector does not require priori knowledge about the frequency locations of the signals of interest. Moreover, it has lower complexity of both samples and computation compared to CD in sparse case. Experimental results show the detector outperforms ED in the sub-Nyquist regime especially in low SNRs.