Stochastic Game Approach for Long-Term Fairness-Aware Radio Resource Allocation

Abstract

Efficient and fair radio resource allocation is essential for wireless networking systems. Most existing fairness-aware dynamic spectrum allocation-based schemes target at instantaneous or short-term resource allocation optimization. However, due to the dynamic nature of wireless networks (in terms of network topologies, traffic demands, qualify-of-service (QoS) requirements, etc.), the resource allocation problem cannot be optimized by only considering instantaneous or short-term resource allocation. In fact, with the fast growing applications of elastic traffic (e.g., video streaming, video conferencing), it is highly desirable to achieve efficient and fair radio resource allocation over a long period of time. In this paper, we propose a long-term fairness-aware dynamic resource allocation mechanism to achieve better system performance (in terms of total network data rate and user fairness). To model the optimal long-term fairness-aware dynamic resource allocation, the stochastic game is adopted to analyze the effect of historical resource allocation on the future network performance. In this scheme, the efficient and fair radio resource allocation for each time slot is optimized with the consideration of the long-term utility. The simulation results show significant performance enhancement in terms of both fairness index and throughput of our mechanism compared to classic dynamic spectrum allocation schemes.