Kolmogorov-Smirnov Test for Spectrum Sensing: From the Statistical Test to Energy Detection

Abstract

Spectrum sensing belongs to important parts of Cognitive Radio (CR) chain. Many different spectrum sensing methods are known. One of the recently proposed approaches to spectrum sensing in cognitive radio systems is based on the Kolmogorov-Smirnov statistical (K-S) test. Statistical K-S test is classified as a non-parametric method to measure the goodness of fit between two distribution functions - the one of the received communication signal and the second of the channel noise. We assume the cumulative distribution function of the noise corresponds to the Additive White Gaussian Noise (AWGN) and is known in advance. The paper discusses two modifications of the Kolmogorov-Smirnov test - the first with the removed information about the signal energy and the second taking it into account for decision. The experimental results prove the robustness of the algorithm for different kinds of received signals.