Analysis and Optimization of Random Sensing Order in Cognitive Radio Networks

Developing an efficient spectrum access policy enables cognitive radios to dramatically increase spectrum utilization while ensuring the predetermined quality of service levels for primary users (PUs). In this paper, the modeling, performance analysis, and optimization of a distributed secondary network with a random sensing order policy are studied. Specifically, secondary users (SUs) create a random order of available channels upon PUs' return, and then, they find optimal transmission and handoff opportunities in a distributed manner. By a Markov chain analysis, the average throughputs of the SUs and the average interference level among the SUs and the PUs are investigated. A maximization of the secondary network performance in terms of the throughput while keeping under control the average interference is proposed. It is shown that, despite traditional views, a nonzero false alarm in the channel sensing can increase channel utilization, particularly in a dense secondary network where the contention is too high. Then, two simple and practical adaptive algorithms are established to optimize the network. The second algorithm follows the variations of the wireless channels in nonstationary conditions and outperforms even static brute force optimization while demanding few computations. The convergence of the distributed algorithms is theoretically investigated based on the analytical performance indicators established by the Markov chain analysis. Finally, numerical results validate the analytical derivations and demonstrate the efficiency of the proposed schemes. It is concluded that fully distributed sensing order algorithms can lead to substantial performance improvements in cognitive radio networks without the need for centralized management or message passing among the users.