Distributed downlink resource allocation in cellular networks through spatial adaptive play

In this work, we develop mathematical and algorithmic tools for distributed resource allocation in downlink of mobile cellular networks. Our algorithms perform power allocation, subcarrier selection and base station association simultaneously. We aim to maximize the aggregate utility of all the users where users' utilities can be arbitrary increasing functions of their throughputs; this allows us to capture both elastic and inelastic traffics. Our solution is via framing the problem as a potential game among users. We propose a highly scalable, asynchronous algorithm that provably converges to a Nash equilibrium of this game. This algorithm requires only local measurements, limited communication between neighboring nodes and limited computation. This algorithm may at times stuck at a local maximum. To alleviate this problem, we propose an enhanced randomized algorithm based on spatial adaptive play that provably converges to a system optimal resource allocation. We also present simulation results to illustrate convergence and performance of the proposed algorithms.