QoS-Aware Channel Selection in Cognitive Radio Networks: A Game-Theoretic Approach

In a cognitive radio wireless network, each node can sense and opportunistically access the under-utilized spectrums in the primary system. Since the unoccupied spectrum is location-dependent and time-dependent, the available spectrums in each node are different. With this spectrum heterogeneity and different Quality-of-Service (QoS) requirement, different nodes may have different preferences in using a particular channel to communicate with its neighboring nodes. In this paper, we formulate this channel selection problem using a cooperative game theoretical approach such that the QoS requirement of each node is guaranteed and the network capacity is maximized. To further improve the performance, a learning negotiation mechanism is introduced. The key motivation is to derive new mixed strategies of the nodes with the reference to the historical profiles of all selected strategies in the past. Simulation results are presented to show the fast convergence of the game, the high efficiency of the learning mechanism, and the effect of mobility.