Adaptive error control based on type-I hybrid-ARQ protocols

An adaptive error control technique based on the use of type-I hybrid-ARQ protocols is presented. Nonstationary channels are modeled as finite-state Markov chains. The system counts the number of retransmission requests during an observation interval called a frame and, together with knowledge of the code in use and a set of thresholds, estimates the state of the channel. The code which maximizes throughput while maintaining a minimum level of reliability is chosen for use in the next frame. The optimal code for each state is determined a priori by varying both the rate and the number of errors corrected by a family of fixed-length codes. The procedure is demonstrated using Reed-Solomon codes