Implementation of blind cyclostationary feature detector for cognitive radios using USRP

Abstract

Cognitive radio is an emerging technology that is used to solve the problem of scarce spectrum resource utilization. Among its fundamental functions, the most important is the spectrum sensing which requires high accuracy and low complexity particularly at very low signal-to-noise ratio (SNR) values. In this paper, we discuss a recently proposed spectrum sensing detector [1] which explores the sparsity of the Cyclic Autocorrelation Function (CAF), and we analyze its complexity and performance using GNU radio and USRP over real radio channel environment. The presented detector exploits the intrinsic symmetry property and the sparse feature of the CAF in the cyclic frequency domain. Unlike the conventional energy detector and the Dandawaté & Giannakis's algorithm, the implemented detector does not need any prior information neither on the noise variance nor on the primary user's signals. Measurements show that the presented detector performs quite well and it has a low sensing-time in comparison to the classical Dandawaté & Giannakis's algorithm.