Computer Science · Computational Theory and Mathematics
Saving memory in turbo-decoders using the Max-Log-MAP algorithm
See on the mapAbstract
The implementation of the MAP (or BCJR, APP...) algorithm requires a backup memory for the probabilities which can be of great size. In this paper, turbo decoding processes are defined which use little memory, while keeping good capacities of correction. The decoding processes were validated by simulation (C, VHDL). The structures presented offer the designer a panorama of technical solutions for implementing a turbo-decoder starting from the max-log-MAP algorithm.