An Improved SC Flip Decoding Algorithm of Polar Codes Based on Genetic Algorithm

• Published in: IEEE access Vol. 8; pp. 222572 - 222583
• Main Authors: Wang, Xiumin, Ma, Qiangqiang, Li, Jun, Zhang, Hongchao, Xu, Wenchao
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Binary system; Candidate flipping positions set; Codes; Complexity; Complexity theory; Decoding; genetic algorithm; Genetic algorithms; Genetics; Mathematical analysis; Network latency; Performance indices; Polar codes; Reliability; Simulation; successive cancellation flip; Wireless communication systems; Wireless communications
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2020.3041290

Polar codes have been applied for physical downlink control channel in the <inline-formula> <tex-math notation="LaTeX">5^{\mathrm {th}} </tex-math></inline-formula> generation wireless communication system. Although successive cancellation flip (SCF) decoding algorithm can improve decoding performance of polar codes, it has also led to the increasing in decoding latency and calculation complexity. Candidate flipping positions set (CFPS) of traditional SCF decoding is consisted of indexes of all information bits. However, some subchannels are reliable enough so that it is almost impossible to cause decoding errors for these subchannels. In order to reduce decoding latency and calculation complexity of SCF decoding algorithm, a new method of constructing the CFPS based on genetic algorithm (GA) is proposed in this paper. What's more, the paper fills a gap of applying GA for decoding of polar codes. In our proposed method, indexes of all information bits are used as individuals of GA. Then through some genetic operations, a vector that can indicate the reliability of all information bits is obtained. Based on the obtained vector, a new CFPS is constructed. Simulation results show that SCF decoding algorithm based on CFPS constructed by GA can achieve competitive decoding performance, while keeping lower calculation complexity and decoding latency. Compared with SCF decoding algorithm based on critical set, the normalized decoding latency of proposed SCF decoding algorithm can be reduced by 39% at 1.5dB when code length and code rate are equal to 1024 and 0.5, respectively.