Single-Minimum LDPC Decoding Offset Optimization Methods

Abstract

Low-density parity-check codes are widely used in communication systems. To meet the high throughput and energy efficiency requirements of current and future systems, it is desirable to further simplify the decoder. Quantized min-sum (MS) decoders are of particular interest for their low implementation complexity, which can be further reduced by computing a single minimum (SM) during check node update, instead of two. However, this simplification can lead to poor decoding performance unless it is carefully incorporated. In this paper, we formalize a general optimization problem for SM decoding, and propose search heuristics to solve it. In addition, we provide density evolution (DE) equations for the first two decoding iterations that properly take into account the lack of extrinsic update rule, and show that this DE result can be used to obtain good solutions to the SM optimization problem with low computational complexity.