Channel coding to exploit soft decisions and error-dependence in channels with memory.

Abstract

Many real-world communications channels exhibit memory. That is, the sequence of symbol errors in the channel output contains a statistical dependence. Frequently the dependence in this error process takes the form of burstiness; channel errors tend to occur in short bursts with long error-free intervals in between. For an application involving random-error correcting codes, this error dependence is typically removed (actually just ignored) by interleaving/deinterleaving. In this thesis we show that with a moderate increase in decoder complexity, soft-decision information and knowledge of the channel memory can be used to improve decoder performance. In this thesis we introduce a new, general, error-correction coding scheme for use on soft-decision channels with memory. Then, for the specific case of a slow-frequency-hopped code-division multiple-access (SFH-CDMA) channel, it will be shown that a significant performance gain can be obtained by using this new decoder. The results of the SFH-CDMA investigation demonstrate that for an SFH-CDMA system appropriate to a cellular telephone network, the new decoder can increase the code rate by 10% to 300% over a baseline decoder using the same interleaving frame but based on hard-decision, random-error correcting coding. This performance advantage requires roughly a three to eight-fold increase in the complexity of the decoder.