Adaptive HARQ With Non-Binary Repetition Coding

We consider Incremental Redundancy Hybrid Automatic Repeat reQuest (IR-HARQ) in which the code rate and modulation of the initial transmission and all retransmissions are adjusted based on average channel statistics. In the absence of instantaneous channel state information at the transmitter (CSIT), we present a method which computes, prior to transmission, the optimum code rates and modulations and explicitly considers a given maximum number of retransmissions. For the case that additional feedback on CSI of previous transmission attempts is available, we present two heuristic schemes which exploit this knowledge and offer increased throughput at the cost of higher computational complexity. We employ a rate-adaptive non-binary LDPC coding scheme which makes use of non-binary repetitions. While this coding scheme is particularly well-suited for adaptive IR-HARQ, we note that the presented analysis can be applied to any other channel code which employs soft decoding.