Output decisions for stochastic LDPC decoders

Abstract

Stochastic decoding is a method to iteratively decode error-correcting codes such as low-density parity-check codes. Due to the computational simplicity of its decoding algorithm, stochastic decoding provides not only low silicon area but also competitive performance. In this paper, we research the different techniques used to determine the code bits generated by stochastic decoders during the final decision process. We investigate an efficient method where code bits are directly decided based on the final stochastic bits generated at output nodes. This observation contributes to saving chip area and providing high throughput for stochastic decoders.