Delay and Jitter Constrained Wireless Scheduling with Near-Optimal Spectral Efficiency

Next generation wireless schedulers will support increasingly heterogeneous devices/applications in terms of their traffic characteristics and service requirements. Particularly challenging is the need to deliver traffic subject to delay and reliability constraints in a spectrally efficient manner. We propose a new measurement-based Opportunistic Guaranteed Deadline Scheduler (OGDS) that meets strict delay deadlines on users’ packets. This is achieved by scheduling packet transmissions when the current channel rate is better than that expected in the time window before packet deadlines expire. In order to meet such requirements one must have a complementary admission control policy. We exhibit a simple, once again measurement based policy, that indirectly accounts for heterogeneity in traffic, channel and delay constraints by monitoring statistics of OGDS’s resource usage. We show via extensive synthetic and trace driven simulations that OGDS requires at most 10−25% more resources compared to an optimal offline scheduling policy with complete knowledge of future channel rates, and performs much better than standard baselines including the state-of-the-art MLWDF scheduler. Finally, we propose a modification to OGDS that enables one to control the jitter at a possible loss in spectral efficiency.