Approximation algorithms for wireless opportunistic spectrum scheduling in cognitive radio networks

Given a set of communication links in cognitive radio networks, assume that the underlying channel state information along each link is unknown; however, we can estimate it by exploiting the feedbacks and evolutions of channel states. Assume time is divided into time-slots. Under the protocol interference model, the opportunistic spectrum scheduling problem aims to select interference-free links to transmit at each time-slot to maximize the average throughput over the long time horizon. Existing works on the opportunistic spectrum scheduling problem cannot satisfyingly address the wireless interference constraints. We apply the framework of restless multi-armed bandit and develop approximation algorithms for the problem with stochastic identical links and nonidentical links respectively. Based on the updated estimations of channel states, the proposed algorithms keep refining future link scheduling decisions. We also obtain approximation bounds of these two proposed algorithms.