Concurrent decoding of turbo codes

Abstract

In conventional decoding of rate-1/3 turbo code, there are two maximum a posteriori(MAP) computing blocks, $DEC_1$ and $DEC_2$ respectively. When the former is on, the latter is idle, and vice versa. In this thesis, we propose a concurrent decoding method which computes two log-likelihood ratios simultaneously at each iteration step from $DEC_1$ and $DEC_2$, and combines them to get a diversity gain. Our results indicate that the required $E_b/N_o$ for the bit error probability of $10^{-3}$ is reduced by about 0.4dB when the number of iteration is 1, and the reduction in $E_b/N_o$ increases as the code rate decreases.