Turbo-like soft-decision decoding of Reed-Solomon codes

We aim to bridge the gap between classical coding theory and soft decoding. Reed-Solomon (RS) codes are chosen as an example, and a few standard possibilities for a soft RS decoder are explored. Soon, it is noticed that the resulting (standard) algorithms are too complex and/or not very performant. However, by employing a special critically sampled filter bank representation for the RS codes, an algorithm is developed that shows a good tradeoff between computational complexity and soft decoding performance. Furthermore, it is demonstrated that this algorithm shows remarkable similarities with turbo codes, thereby preserving the perfectness of the original RS code.