Multi user Opportunistic Spectrum Access in Cognitive Radio Networks: An Optimal Stopping Approach with Spectrum Partition

This paper studies a multiuser optimal stopping strategy for Opportunistic Spectrum Access (OSA). Since all secondary users (SUs) share the same spectrum information and use the same OSA strategy to seize available spectrum resources, collisions will happen inevitably. We formulate this problem as an optimal stopping problem and propose an effective approach named "Spectrum Partition" to avoid such collisions. Spectrum is partitioned completely into several channel sets, the amount of which equals to that of SUs. Thus, the problem can be simplified into multiple single-user cases. In each slot, each SU acquires channel sensing order in descending the idle probabilities of channels based on the updated belief vectors and utilizes optimal stopping rule to determine which channel to access in consideration of channel states and exploration overhead. Simulation results are provided to show the effectiveness of the proposed strategy.