Dynamic control of tunable sub-optimal algorithms for scheduling of time-varying wireless networks

It is well known that the Generalized Max-Weight Matching (GMWM) scheduling policy, and in general throughput-optimal scheduling policies, often require the solution of a complex optimization problem, making their implementation prohibitively difficult in practice. This has motivated many researchers to develop distributed sub-optimal algorithms that approximate the GMWM policy. One major assumption commonly shared in this context is that the time required to find an appropriate schedule vector is negligible compared to the length of a timeslot. This assumption may not be accurate as the time to find schedule vectors usually increases polynomially with the network size. On the other hand, we intuitively expect that for many sub-optimal algorithms, the schedule vector found becomes a better estimate of the one returned by the GMWM policy as more time is given to the algorithm. We thus, in this paper, consider the problem of scheduling from a new perspective through which we carefully incorporate channel variations and time-efficiency of sub-optimal algorithms into the scheduler design. Specifically, we propose a Dynamic Control Policy (DCP) that works on top of a given sub-optimal algorithm, and dynamically but in a large time-scale adjusts the time given to the algorithm according to queue backlog and channel correlations. This policy does not require the knowledge of the structure of the given sub-optimal algorithm, and with low-overhead can be implemented in a distributed manner. Using a novel Lyapunov analysis, we characterize the stability region induced by DCP, and show that our characterization can be tight. We also show that the stability region of DCP is at least as large as the one for any other static policy. Finally, we provide two case studies to gain further intuition into the performance of DCP.