Web proceedings papers


Dejan Spasov


Given a M-state convolutional encoder, computing the alfa probabilities of the MAP decoding algorithm can be done with 2 M memory elements, at the cost of increased time complexity. Our aim in this paper is to keep the time complexity unchanged and to develop an algorithm that lowers the memory requirements of the decoder. Our experiments with rate- 1 2 1025 -bit-long Turbo Codes show space reduction for 50% for the MAP algorithm while losing less than 0.2 dB of the error correcting capability of Turbo Codes.


MAP decoding; MAX-Log-MAP decoding; turbo codes; convolutional codes.