Low-Complexity Construction of Polar Codes Based on Genetic Algorithm

Abstract

How to efficiently rank the reliability order of the channels is critical for polar code construction. A new polar code construction framework tailored to a given decoding algorithm based on the genetic algorithm has been proposed to achieve an error correction performance gain. However, the time complexity of this scheme is extremely high. In this letter, a scheme based on a hash table is proposed to generally reduce the computational complexity of the repeated candidates in the genetic algorithm-based system. A two-step method is proposed to improve the crossover process when the system has prior knowledge. It is shown from the numeric results that the proposed scheme achieves computational complexity reduction of the repeated candidates. An error correction performance gain is achieved over the original scheme under successive cancellation list decoding.