A repeat request strategy based on sliding window decoding of convolutional codes

We investigate a decision feedback strategy for convolutional codes which is based on a sliding window decoding procedure and a threshold test as decision rule. For this purpose, we introduce the burst distance spectrum of a convolutional code and derive asymptotic bounds for the ensemble of periodically time-varying convolutional codes. These results are helpful for the asymptotic analysis of the decision feedback scheme. Unit memory codes are particularly suited for such a transmission scheme. For these codes, the decoding procedure is reduced to the decoding of block codes with lengths in the order of the overall constraint length of the convolutional code. This leads to a significantly smaller decoding complexity compared with other known decision rules. Whereas the achievable asymptotic performance is close to the best known bounds. For low rates, our results even improve these bounds.