Neural network decoding of turbo codes

Berrou et al. (1993) presented a new class of codes (turbo codes) whose performances in terms of bit error rate (BER) are very close to Shannon's limit. Very low BER of 10/sup -5/ at E/sub b//N/sub o/ ratio of 0.7 dB has been achieved using iterative decoding of turbo codes. However, the decoding scheme is characterized by complex receiver structures, high computational costs, and long iterative procedures. In this paper, we describe a neural network decoder for turbo codes. Simulation results reveal that this novel decoder has a performance very close to the maximum a posteriori decoder.