Multi-phase predictive proportional fairness scheduling for a multi-channel wireless packet system

In this paper, we propose a practically realizable version of a proportional fairness scheduling algorithm that is applicable to emerging broadband wireless access systems with multiple channels, whose individual channel quality vary for every user, e.g., as in the band AMC (adaptive modulation and coding) mode of a subcarrier allocation in an OFDMA (orthogonal frequency division multiplexing) system. In order to circumvent the complexity of the optimal multi-channel PF algorithm (which grows exponentially with the number of channels), we introduce the notion of the multiphase predictive approach which takes the channel conditions over the multiple slots into account, as opposed to the conventional multi-channel PF scheduling algorithm which deals with the channels in each time slot. When a simple channel-by-channel scheduling algorithm is implemented with the proposed approach, it has been shown that the network utilization with respect to proportional fairness can be further improved over the conventional multi-channel PF scheduling algorithm while resorting to much less complexity. We also present specific examples of implementation for the proposed approaches, along with their performance.¹