부산대학교 도서관

In this paper, we focus on arbitration inter-frame space (AIFS), an essential mechanism within the enhanced distributed channel access (EDCA) to support service differentiation. A novel modeling approach is proposed for throughput optimization and service differentiation guarantee of EDCA-based networks with AIFS differentiation. In particular, a discretetime Markov renewal process is established to model the channel state transition of EDCA-based networks, and the channel state is segmented into more granular states according to AIFS differentiation. Based on this approach, explicit expressions of the throughput of each access category (AC) are obtained as functions of the transmission probability and AIFS number. An optimal transmission probability and AIFS number to maximize the total network throughput under target service differentiation, i.e., a given ratio of node throughput of ACs, are derived. Simulation results demonstrate that with the optimal setting derived in this paper, both the maximum total network throughput and the extreme target throughput ratio can be achieved.