Error-rate analysis of polar code designs using genetic algorithm for additive white Gaussian noise channel

A novel scheme for designing polar codes with specific decoding schemes in the additive white Gaussian noise channel is presented in this paper. The code construction strategy is built on the genetic algorithm (GA) where successive evolution of populations (or group of information sets) leads toward fittest candidate to attain finest error-rates. In this work, it is shown that better error-rates can be attained by both successive cancelation list decoding using GA with no cyclic redundancy check (CRC) and belief propagation decoding using GA with no CRC and no list, compared to existing polar decoding schemes. Our proposed polar code design scheme using GA has the ability to attain a target block error rates with the least possible SNR and using no additional CRC by exploiting least belief propagation iterations or lesser successive cancelation list list size with no self-alterations in the decoding algorithm. This paper provides a novel polar codes design scheme using genetic algorithm (GA) for additive white Gaussian noise (AWGN) channel. Improved block error rates (BLER) can be attained by both successive cancelation list (SCL) and belief-propagation-list (BPL) decoding using GA with no cyclic redundancy check (CRC), compared to existing polar decoding schemes. Proposed scheme attains a target BLER with the least possible signal-to-noise ratio (SNR) and without any additional CRC by exploiting least BP iterations or lower list size in SCL without any self-alterations in the decoding algorithm.image