Spatially-coupled low-density parity check codes: Zigzag-window decoding and code-family design considerations

Abstract

Spatially-coupled low-density parity-check (SC-LDPC) codes are new capacity-achieving codes, which combine the advantages of both convolutional-turbo codes and LDPC block codes. SC-LDPC codes can provide scalable code-length, high granularity, convolutional-gain, low-complexity decoding, and high parallel processing. To overcome the limitations of the current SC-LDPC decoders which have essentially hindered the widespread adoption of SC-LDPC, we propose a new decoder which we call the Zigzag Decoder, that essentially leverages the convolutional-gain of the sliding window decoder. We show that the zigzag-window decoder reduces the average number of iterations and improves the performance relative to the conventional sliding-window decoder. Furthermore, for facilitating the use of SC-LDPC codes in practice, we have explored how to generate SC-LDPC code families and compare the performance and complexity between shortening-based and puncturing-based SC-LDPC code families supporting different code rates.