Performance analysis of spectrum sensing under coloured noise

Abstract

Spectrum sensing (SS) is an essential requirement in Cognitive Radio (CR) to detect primary user (PU) and to access the opportunistic spectrum for secondary users. Several sensing techniques are limited by multipath fading and shadowing which degrade the sensing performance. Hence noise plays an important role in spectrum sensing. Herein, we analyze the condition of correlated noise based on eigenvalue technique. We consider Standard Condition Number (SCN) based statistics for decision that statistics are further based on Random Matrix Theory (RMT). First we analyze the eigenvalue based Marchenko-Pastur (MP) Law in presence of correlated noise. Due to degradation in performance of MP Law, we use new SCN based threshold. We analyze that the new bound increases the performance in case of correlated noise.