Losing Opportunism: Evaluating Service Integration in an Opportunistic Wireless System

In this paper we evaluate interactions among flow-level performance metrics when integrating QoS and best effort flows in a wireless system using opportunistic scheduling. We introduce a simple flow-level model capturing the salient features of bandwidth sharing for an opportunistic scheduler which ensures a mean throughput to each QoS stream for every time slot. We then explore the flow-level performance showing that integration of QoS and best effort flows results in loss in opportunism, which in turn results in a reduction of the stability region, degradation in system throughput, and increased file transfer delay. These losses are shown to be proportional to opportunistic gains, the guaranteed bandwidth and number of QoS flows, but inversely proportional to SNR under a Rayleigh fading channel model. In an integrated system exploiting opportunism, local instability appears to be more severe than in wired networks and average delays experienced by best effort flows are prolonged. We suggest that a form of admission control for best effort flows is necessary to avoid local instability, and ensure adequate performance.