Throughput Optimal Multi-Slot Sensing Procedure for a Cognitive Radio

We consider a cognitive radio system with M primary channels where users' data transmissions follow a slotted structure. We consider the scenario where the channel availability statistics are correlated across channels as well as across time slots. In each slot, the cognitive user (CU) senses channels one by one until a suitable channel is found for data transmission. CU adapts theCognitive Radio data rate based on the channel's fading gain. We employ a Markov chain model to capture the channel availability statistics across the slots. We address the problem of finding optimal sensing order/procedure in every slot based on the history of sensed channels (including the previous slots), in order to maximize the total CU throughput over N-slots. Using theory of optimal stopping, we derive recursive expressions for the optimal multi-slot CU throughput and find the optimal multi-slot sensing order/procedure using dynamic programming. We also study few sub-optimal sensing procedures. Using numerical results, we illustrate the gains in exploiting the correlation of channel availability statistics.