Error patterns in belief propagation decoding of polar codes and their mitigation methods

Abstract

Belief propagation (BP) is a high-throughput decoding algorithm for polar codes, but it is known to underperform successive cancellation (SC) decoding and list decoding in error-correcting performance. In this work, we study the error patterns of BP decoding of polar codes to uncover the error mechanisms, as well as the influence of channel condition, code design, and decoder implementation. Based on the insights, we design new ways to detect, prevent and overcome BP decoding errors.