Optimising the activation cycle of Turbo Decoders using three-dimensional extrinsic information transfer charts

In this study, activation cycle of turbo detection system is investigated using three-dimensional extrinsic information transfer charts. Turbo detection typically employs an iterative process of information transfer with a fixed activation schedule. In the proposed work, the activation cycle of different component decoders is adapted for each signal-to-noise ratio value using information obtained from extrinsic information transfer charts. The objective is to reach the desired reliability of estimates at a lower complexity and for that purpose an exhaustive tree search-based message passing algorithm has been adopted. The results are validated by showing that the optimum activation cycles are similar to that of fixed scheduler with matched component decoders. The resulting bit error rate curves closely approach and in some cases excel the results obtained for the fixed activation while operating at only a fraction of their complexity.