Multistage Wiener filter aided MDL approach for wideband spectrum sensing in cognitive radio networks

Abstract

Wideband spectrum sensing in cognitive radio networks has gained significant research interests in the recent years. In the context of noise uncertainty, the noise variance has no access to accurate estimation which brings about an imprecise decision threshold. This paper investigates this subject from a novel perspective. Multistage Wiener filter (MSWF) is adopted in the conventional minimum description length (MDL) criterion to enhance the detection performance. By seeking the value that minimizes the MSWF aided MDL criterion, the number of occupied sub-channels in the wideband spectrum is determined. And then the accurate locations can be obtained based on the received energy of each channel. The proposed scheme is robust to noise uncertainty since it requires no estimation of noise variance. Meanwhile, having no demand for the estimation of covariance matrix or its eigenvalue decomposition makes our approach computational attractive. In addition, the proposal partitions the array data into the cleaner signal and noise subspace components and can thereby improve the detection performance. Numerical results verify our approach and show that it is superior to other existing sensing algorithms in the previous works.