Rate Selection Algorithms for IR Hybrid ARQ

We formulate the N-stage problem of Adaptive Incremental Redundancy (IR) Hybrid ARQ (HARQ) rate selection given delayed channel mutual information feedback and fixed packet size as a Markov Decision Process over a finite horizon. We obtain a conditional throughput maximizing rate allocation algorithm under a conditional outage probability constraint given an initial state and numerically evaluate its performance for a quantization of the state and control space for the case of N = 2.