Review and Evaluation of Belief Propagation Decoders for Polar Codes

The polar code has become one of the most popular and important forward error correction (FEC) coding schemes due to its symmetric characteristics of channel polarization. This paper reviews various decoding schemes for polar codes and discusses their advantages and disadvantages. After reviewing the existing performance-enhancing techniques such as belief propagation decoding with list, a new method is proposed to further improve the performance. In addition, a new complexity reduction technique based on the constituent codes is proposed, and a new scheduling scheme is introduced to reduce the decoding latency. Due to the recent development of neural networks, their applications to decoding schemes are also reviewed and evaluated. Finally, the proposed complexity-reduced technique is integrated with a neural network-based belief propagation decoding, which demonstrates performance enhancement as well as computational complexity reduction.