Two Efficient Stopping Criteria for Iterative Decoding

The paper proposes two new stopping criteria for iterative decoding of turbo codes. The first criterion is based on the mutual information between the logarithm likelihood ratio and the data bits, and the second one is based on the a posteriori error probability of the decoded information bits. Compared to the existing stopping rules, the proposed criteria need not to store any extra data, and can reduce the average number of iterations much closer to the genie stopping rule when the performance degradation is negligible.