Decoding inter-block permuted turbo codes based on multi-stage factor graphs

Abstract:
Inter-block permuted turbo codes (IBPTC) provide significant performance gain over the conventional turbo codes (TC) without increasing the encoding and average decoding latencies. The major shortcoming is the increased memory requirement. A new decoding schedule is proposed to reduce both the memory requirement and the decoding latency. We present a multi-stage factor graph and sub-graph approach to describe the schedule and message-passing flow of an IBPTC decoder. Evaluating the effect of the finite iteration number on the decoder performance becomes feasible through the use of such graphs. The proposed graphic approach can be applied to find more efficient (in terms of computing complexity and storage) decoding schedules for other iterative decoders as well.