On channel-adaptive fair multiple access control

Abstract

Multiple access control (MAC) of the uplink in a wireless mobile computing system is one of the most important resource allocation problems in that the response time and throughput of user applications (e.g., wireless Web surfing) are critically affected by the efficiency of the MAC protocol. Compared with a traditional MAC problem (e.g., the wireline Ethernet), there are two important new challenges in a modern wireless network: (1) multimedia data with diverse traffic requirements are involved; and (2) the wireless channel has a time-varying quality for each user. Furthermore, a more prominent user requirement is fairness among different users, possibly with different traffic demands. While some protocols have been suggested to handle multimedia data and/or tackling the time-varying channel, there are a number of drawbacks in these existing protocols. The most notable drawback is that the channel model is rather unrealistic - just using a two state Markov chain instead of relying on accurate models of multipath fading and shadowing effects. Another common deficiency is that fairness is ignored. In this paper, we propose to use a new notion of fairness that can capture a realistic channel model, and to integrate a fair queueing scheduling algorithm in a MAC protocol to optimize performance while maintaing fairness among users regardless of their channel states and data types.