State-dependent proportional fair scheduling algorithms for wireless forward link data services

For wireless forward link data services, the proportional fair algorithm (PFA) incorporating opportunistic scheduling was originally designed to maintain service fairness among users as well as to exploit channel fluctuations in fading environments. The operation of PFA necessitates throughput monitoring for each user, in general through exponentially smoothing received services. In the paper, we adapt the PFA for use in more generic traffic conditions where the number of active users may change with time and data for a user may arrive intermittently. Our approach is not only to simply stop throughput monitoring for idle users but also to further accommodate in throughput monitoring changes in the number of active users. Moreover, we use in throughput monitoring different smoothing factors depending on the state of backlogged data loads, in order to exploit the variation of multiuser diversity gain. Consequently, our PFA, taking into account both channel state and system queue fluctuations, presents several favorable performance features.