Performance tradeoffs in wireless scheduling with flow aggregration

We develop Markov models for various schedulers in order to evaluate their performance tradeoffs at wireless links. While FIFO scheduling aggregates flows into a single flow just prior to the wireless link, channel-state dependent (or CSD) schedulers maintain a queue for each flow, and use predicted channel information to make scheduling decisions. We present results of tradeoffs in terms of overall throughput and per-flow QoS performance obtained with each scheduler under different channel conditions.