Hybrid Decoding of Finite Geometry LDPC Codes

TL;DR

This paper introduces hybrid decoding schemes for finite geometry LDPC codes that combine bit flipping and Min-Sum algorithms, significantly reducing computational complexity without sacrificing performance.

Contribution

It proposes a novel hybrid decoding approach that balances complexity and performance for finite geometry LDPC codes, supported by simulation and complexity analysis.

Findings

Hybrid schemes reduce decoding complexity substantially.

Most decoding load is handled by the low-complexity BF variant.

Performance and convergence are maintained comparable to MS decoding.

Abstract

For finite geometry low-density parity-check codes, heavy row and column weights in their parity check matrix make the decoding with even Min-Sum (MS) variants computationally expensive. To alleviate it, we present a class of hybrid schemes by concatenating a parallel bit flipping (BF) variant with an Min-Sum (MS) variant. In most SNR region of interest, without compromising performance or convergence rate, simulation results show that the proposed hybrid schemes can save substantial computational complexity with respect to MS variant decoding alone. Specifically, the BF variant, with much less computational complexity, bears most decoding load before resorting to MS variant. Computational and hardware complexity is also elaborated to justify the feasibility of the hybrid schemes.