Adaptive multiple-transmission block error control coding based on a minimal hidden Markov model of the HF error channel

Presents a new algorithm for estimating parameters in a 2-binary symmetric channel-state hidden Markov model for the HF radio channel, based on a 2D optimisation by binary search of "bad" state parameters. Excellent modelling accuracy has been obtained, as shown by close fits between distributions of various error statistics for actual and simulated data. The performance of the multiple transmission scheme without FEC and without adaptivity is not impressive. FEC alone and adaptivity alone offer some advantages, but these seem too small to justify the increase in complexity. The best performance is obtained when both FEC and adaptivity are employed: with triple-error correction, single-repetition in the "good" state and triple-repetition in the "bad" state, a respectable block throughput of 0.69 (bit throughput of 0.36) is achieved.<
>