Heuristic Tree Search for Detection and Decoding of Uncoded and Linear Block Coded Communication Systems

A heuristic tree search algorithm is developed for the maximum likelihood detection and decoding problem in a general communication system. We propose several "cheap" heuristic functions using constrained linear detectors and the minimum mean square errors (MMSE) detector. Even though the MMSE heuristic function does not guarantee the optimal solution, it has a negligible performance loss and provides a good complexity-performance tradeoff. For linear block coded systems, heuristic tree search is modified for soft decision decoding. High rate codes are decoded via the minimum state trellis, and low rate codes via the minimum complexity tree. Preprocessing is also discussed to further speed up the algorithms.