MDP-Based MAC Design With Deterministic Backoffs in Virtualized 802.11 WLANs

摘要

This paper presents medium access control (MAC) protocols for a virtualized 802.11 network aiming to improve network performance and isolation among service providers (SPs). Taking into account the statistical properties of arrival traffic, a Markov decision process (MDP) is formulated to maximize the network throughput subject to SP reservations. By introducing the policy tree of the MDP, we present an optimal access policy. Each user can track this policy tree by carrier sensing and learn its transmission opportunity. As the computational complexity of the policy tree exponentially grows with the total number of users, an efficient heuristic algorithm is proposed based on the MDP formulation, where each user is assigned a deterministic backoff value. Numerical results show that the performance of the proposed heuristic algorithm closely matches that of the optimal policy. Moreover, both optimal and heuristic algorithms significantly improve time-division multiple access (TDMA) and carrier-sense multiple access (CSMA) in terms of packet delivery ratio and isolation in unsaturated networks.