On the Performance of SC-Flip and SC-Perturbation Decoders for Parallel Decoding of Polar Codes

In ultra-high frequency band communication for 5G or beyond 5G, high-speed transmission of the Tbps-level is targeted [1]. To reach that level, parallelization of decoding algorithms and architecture is crucial. There have been studies that have implemented the high-speed polar decoder, but it was possible to implement only under compromise in the error-rate performance, because of the use of successive cancellation (SC) decoding [2] instead of SC-list (SCL) [3] decoding. Therefore, it is essential to study decoder with improved performance whose performance can be comparable to the SCL decoder while using the SC decoder. The SC-Flip (SCF) [4]–[6] and SC-Perturbation (SCP) decoding [7], [8] as post-processing decoding techniques, improve the performance through re-decoding when the initial SC decoding fails. As the signal-to-noise ratio (SNR) increases, the average complexity converges to the SC decoding level while attaining better error-rate performance than the SC decoder. However, the performance of this post-processing schemes has not been compared nor extensively investigated. Therefore, we analyze the performance of SCF and SCP decoders with respect to the maximum number of decoding iterations for various coding parameters. We confirm that the SCF is better when the maximum number of decoding iterations is small, and the SCP outperforms the SCF as the number of iteration increases.