Switch Packet Arbitration via Queue-Learning

In packet switches, packets queue at switch inputs and contend for outputs. Thecontention arbitration policy directly affects switch performance. Thebest policy depends on the current state of the switch and current traffic patterns. This problem is hard because the state space, possible transitions, and set of actions all grow exponentially with the size of the switch. We present a reinforcement learning formulation of the problem that decomposes the value function into many small independent valuefunctions and enables an efficient action selection.