Reinforcement learning in hierarchical cognitive radio wireless networks

Opportunistic networks are an emerging networking paradigm aiming to exploit the spectrum availability in a distributed ad-hoc manner. In such types of networks communication between source and destination is established on-the-fly and depends on a number of parameters related to the channel. In this paper, we present an algorithm for routing optimization in hierarchical cognitive radio enabled networks, using the access base stations. We describe how spatial and temporal system parameters between nodes can be employed to design optimum routes between the nodes thus becoming invaluable for deriving optimum opportunistic algorithms. Initial results of this work indicate that traffic history can improve the performance of the routing algorithm by identifying the nodes that are most likely to be available for routing thus minimizing retransmissions and reducing blocking probability. One of the challenges in dealing with these records is the memory and processing requirements needed by the power hungry algorithms.