Skewness and pseudocodewords in iterative decoding

Abstract

Impressive performances by low-density parity-check codes and turbo codes have inspired researchers to determine how their iterative decoders depart from optimal decoding. Iterative decoding can be viewed as the application of the sum-product or min-sum algorithm in a graph that describes the constraints on a system of variables. The authors discuss here how iterative decoding maximizes correlation, and skewness and pseudocodewords in the decoding tree.