Due to interference and channel fading, wireless environments suffer large and location dependent error rates which greatly degrade the performance of multimedia applications such as video streaming and so. In wireless LAN (WLAN) such errors can be more dramatic and because the physical layer of WLAN implements its own technique to combat these errors - viz., link adaptation techniques such as the well known auto-rate fall-back (ARF) algorithm - the medium access control (MAC)layer throughput of individual stations is limited by that of the station that experiences the worst channel quality. MAC scheduling is one solution to this problem, and extensions of fair queueing policies have been proposed for wireless environments (e.g., [3, 2] and references therein). Most algorithms extend the classic generalized processor sharing algorithm with an opportunistic channel access. Such opportunistic scheduling relies on a channel prediction model in which the channel is assumed to be a two state (good and bad) Markovian channel. Based on future channel states predictions, transmission opportunities are swapped between stations to allow only stations with good channels to transmit. This has a dramatic impact on delay.