Adaptive scheduling and coding for time-varying links

In dealing with wireless link variability, adaptive scheduling and adaptive coding each has its own advantages and disadvantages. These two approaches were mostly used separately. Existing work on wireless scheduling usually uses a coarse two-stage link model and assumes perfect link knowledge, while work on adaptive coding ignores bandwidth sharing among users. In reality, multiple users with time-varying and location dependent links share the transmission bandwidth in a wireless network, and the two approaches need to be integrated to achieve better utilization and finer fairness granularity. In this paper, we design a simple scheme that combines the adaptive scheduling and coding mechanisms. The scheme is applicable to general multiple-state Markov link model and enables tradeoff between bandwidth utilization and fairness granularity. Link state estimation and prediction errors are accounted for in decision-making. We also show that the optimal solution for maximizing bandwidth utilization is a special case of the general schemes.