Distributed queueing games in interference-limited wireless networks

We study distributed queueing games in interference-limited ad-hoc wireless networks. We formulate system design as a Nash Equilibrium (NE) problem, where the users aim at maximizing their own throughput by choosing the optimal transmission threshold. We first derive conditions for the existence and uniqueness of the NE; then we propose a distributed best-response algorithm solving the game along with its convergence properties. A second contribution of the paper is to develop a Branch and Bound-based (centralized) algorithm solving the associated (nonconvex) social problem, which one can use as benchmark to evaluate the performance of the proposed game theoretical formulation. Interestingly, our numerical results show that the sum-throughput achievable at the NE of the proposed game are very close to that of the social problem, which validates our game theoretical formulation. The performance loss is not negligible only in high interference scenarios. For such cases, we proposed a pricing-based algorithm yielding sum-throughput solutions very close to the globally optimal ones, at the cost of very limited signaling among the users.