Lower-Complexity Layered Belief-Propagation Decoding of LDPC Codes

Abstract— The design of LDPC decoders with low complexity, high throughput, and good performance is a critical task. A well-known strategy is to design structured codes such as quasi-cyclic LDPC (QC-LDPC) that allow partiallyparallel decoders. Also, several works show that sequential schedules, such as Layered Belief-Propagation (LBP), converge faster than the traditional flooding schedule. In this paper, we propose a novel low-complexity sequential schedule called Zigzag LBP (Z-LBP). Current LBP schedules do not allow partially-parallel architectures for some codes, such as high-rate codes with small-to-medium blocklengths. Our proposed algorithm can still be implemented in a partially-parallel manner in these codes. Z-LBP provides the same benefits of LBP such as faster convergence speed and achieves lower frame error rates than flooding.

• Type: Conference
• Year: 2008
• Where: ICC
• Authors: Yuan-Mao Chang, Andres I. Vila Casado, Mau-Chung Frank Chang, Richard D. Wesel
• Keywords: Communications, Current LBP Schedules, ICC 2008, Sequential Schedule, Zigzag LBP