Primary user detection in cognitive radio networks through quickest detection

Abstract

Spectrum sensing is one of the key functionalities in cognitive radio (CR) which enables opportunistic spectrum access. How to promptly sense the presence of the primary user (PU) is a key issue to CR network. The quick detection of the PU is critical such that violation of the detection on time will cause an interference to the PU. Particularly the spectrum sensing can be evaluated by how quickly it detect the changes. This involves detecting reliably and quickly possibly weak primary user signal. In this work, we propose a cluster based quickest change detection algorithm for the spectrum sensing. The Proposed scheme utilizes one-sided CUSUM (cumulative sum) algorithm at the user level, and proposed an algorithm at the cluster level which maintain 1-slot history of the user at the cluster head (CH), which also helps in detecting the change at the cluster level. CH only collect data from the users which observe change in their slot. By exploiting the advantage of one-sided CUSUM algorithm with cluster-based approach, the proposed scheme achieve the minimum detection delay subject to false alarm constraint. With simulation results, we demonstrate that the proposed scheme has faster detection as compare to conventional detection scheme.