A Markov error modeling approach to distributed turbo coding

This paper reexamines the decoding problem for a distributed Turbo code (DTC) over a 3-node relay network applying the decode-and-forward strategy under the half-duplex constraint. The conventional decoder simply assumes that the relay performs error-free decoding. In practice, unsuccessful decoding at the relay causes error propagation, which may become a performance bottleneck. Improved decoders reported in the literature are all based on memoryless error models, although relay decoding errors are actually correlated. In this paper, we propose a new Markov error modeling approach to accurately characterize the memory effect of relay decoding errors. Based on the proposed error model, we derive a new iterative decoder which includes a component decoder implemented with a product trellis structure. It is empirically verified that the new decoder can outperform existing decoders, and can achieve promising performance improvement in the scenario of symmetric links.