Combined wideband and narrowband signal detection for spectrum sensing

Abstract

In spectrum sensing, the goal is to detect the presence of the primary user and to avoid the spectrum utilized by it. The localization algorithm based on double-thresholding (LAD) is a recent method for detecting and localizing signals without a priori knowledge of the noise power. The LAD method performs the best when detecting narrowband (NB) signals. For complete spectrum sensing, detecting just the NB signals is not enough, as the primary user's signal may equally well be a wideband (WB) signal. We propose combining the LAD method with the maximum-minimum eigenvalue (MME) method. The results indicate that the combined method is able to detect both the NB and WB signals without knowledge of the noise power.