Joint Distributed Sensing and Channel Access in Cognitive Radio Networks

We consider a Cognitive Radio Network having one Primary User (PU) and N Secondary Users (SUs). In this paper, we study the problem of joint channel–sensing and channel– access for SUs. When the channel is in use by the PU, the signal that the PU sends and the channel fading gains are unknown to SUs. The channel sensing problem that we consider is detecting whether or not there is an unknown signal (with random fading) in noise. For this channel–sensing problem, we propose a sequential detection procedure based on the energy of samples that each SU observes. As soon as an SU detects the idle/busy state of the channel, it broadcasts it’s local decision to all other SUs. We propose a global decision rule that makes a decision that the channel is idle, only if at least $\Gamma$ out of N SUs have broadcast idle local decisions; otherwise, the global decision rule makes a decision that the channel is busy. Also, the channel access is provided to the SU that is the first one to broadcast an idle decision. We study the detection and false-alarm performance of our proposed procedure, and compare the performance with that of Sequential Probability Ratio Test (SPRT) based sensing procedure. From the Receiver Operating Characteristic (ROC), and the average sample number (ASN) metrics, we observe that our energy based sequential sensing procedure yields a better probability of detection than the SPRT based procedure for a given probability of false-alarm. Also, as the threshold on the number of idle local decisions $\Gamma$ increases, probability of detection also increases, but at the cost of detection delay.