Compressive spectrum sensing using complementary matrices for cooperative cognitive radio networks under a non-reconstruction framework

In this paper, complementary sensing matrices based on Golay-Hadamard codes are proposed for eigenvalue-based spectrum sensing under a compressive, non-reconstruction framework in collaborative cognitive radio networks. The non-reconstruction framework, which is based on random matrix theory, means that compressive sensing is used but the spectrum occupancy can be determined without the need for a reconstruction algorithm. Existing methods in literature rely on the use of identical matrices among the members of a collaborative network for this method to work, which is - in most cases - impractical, particularly in case of the commonly used random sensing matrices. The proposed method uses a set of complementary matrices that can be used by the members of the collaborative network which, when combined, provide accurate sensing results comparable to those obtained when sampling at the Nyquist rate. The proposed method is tested through simulations, and different scenarios are compared to both the non-compressive case and the existing compressive sensing method with identical matrices under additive white Gaussian noise, Rayleigh flat fading and multipath fading. (C) 2019 Elsevier B.V. All rights reserved.