Demand-Aware Opportunistic Spectrum Access: A Game-Theoretic Learning Approach

We study the problem of opportunistic spectrum access with users’ demands in distributed systems. The users’ demands play an important role in estimating the final access results. For example, the same throughput may lead to completely different experience for the users with different demands. To emphasize the influence of the demand, we use the ratio of demand and throughput to consider them together. We focus on the sum ratios of each user to make the resource allocation efficient from the system view. We model the channel selection problem with demand-throughput ratio as a cooperative game, propose an ordered best response algorithm to achieve NE point and prove the existence of NE point. The stochastic learning algorithms has been used in simulations. The results show that the ordered best response algorithm and stochastic learning approach both converged and achieved good performance in fairness and utility which are better than random access situation. what’s more, the ordered best response algorithm has a significant improvement in convergence time.