Constructing Rate-Compatible LDPC Codes with a Novel Efficient Ranking Criterion

Abstract

In this paper, we consider the construction of efficient rate-compatible (RC) low-density parity-check (LDPC) codes. Specifically, we introduce a novel criterion to rank puncturing patterns and splitting patterns for RC LDPC codes. Based on the Gaussian approximation density evolution (GADE), cost functions are devised to characterize the degree distribution of the punctured or split code matrices, which are derived from a given code's parity check matrix by some matrix row transformations. These cost functions allow us to effectively compare the estimated performance of the candidate patterns and to sort out good ones. Based on the proposed ranking criterion, we provide a good RC LDPC code structure, which can support a wide range of code rates and outperform most of the existing RC LDPC codes in literatures. Moreover, the proposed scheme has the important advantage of reduced decoding latency and complexity without sacrificing performance. Numerical results demonstrate that the proposed RC LDPC codes can lead to nearly 25% decoding complexity saving and latency reduction at comparable or even better performance.