Generic Direct Approach for Decoding Turbo Codes Using Probability Density Based Reliability Model

TLDR: In this paper, the reliability has been identified in terms of probability densities of the bit values and the same used with the direct decoding algorithm for turbo codes proposed recently and has achieved a significant coding gain improvement.

Abstract: The probability of the received bit values has been directly used in the performance enhanced reliability based direct decoding algorithm for turbo codes proposed recently and has achieved a significant coding gain improvement. This is in contrast to the maximum a posteriori (MAP)—type decoders and their variants in vogue which use log-likelihood ratio (LLR) in place of reliability and go through an iterative process. In this paper reliability has been identified in terms of probability densities of the bit values and the same used with the direct decoding algorithm. Extensive simulations with commonly used encoders show a clear performance improvement especially for short block lengths. Neither does the approach constrain noise distribution in any way as brought out by considering different distributions.