Recursive enhancement of intrinsic soft information for complex reed-solomon codes

Complex Reed-Solomon codes recently gained attention in deterministic Compressed Sensing (CS) schemes. This allows to exploit ideas and algorithms from algebraic coding theory for reconstruction or recovery in CS. We show how intrinsic soft decision information can be recursively enhanced, allowing decoding/recovery beyond the conventional capabilities. This recursive enhancement is based on error/erasure decoding and improves the soft information. Its performance is evaluated by numerical simulations and compared to previous results and to the performance of Orthogonal Matching Pursuit in combination with optimized matrices.