QoE Driven Decentralized Spectrum Sharing in 5G Networks: Potential Game Approach

This paper studies spectrum sharing for providing better quality of experience in 5G networks, which are characterized by multidimensional heterogeneity in terms of spectrum, cells, and user requirements. Specifically, spectrum access, power allocation, and user scheduling are jointly investigated and an optimization problem is formulated with the objective of maximizing the users' satisfaction across the network. In order to reduce the complexity and overhead, decentralized solutions with local information are required. To this end, we employ game-theoretic approach and interference graph to solve the problem. The proposed game is proved to have at least one Nash Equilibrium (NE), corresponding to either the globally or locally optimal solution to the original optimization problem. A concurrent best-response iterative algorithm is first devised to find the solution, which can converge to an NE, but may not be globally optimal. Therefore, a spatial adaptive play iterative (SAPI) learning algorithm is further proposed to search the global optimum. Theoretical analysis demonstrates that the SAPI algorithm can guarantee to find the globally optimal solution with an arbitrary large probability, when the learning step is set to be sufficiently large. Simulation results are provided to validate the performance of the proposed algorithms.