Decoding LDPC codes via Noisy Gradient Descent Bit-Flipping with Re-Decoding
Decoding LDPC codes via Noisy Gradient Descent Bit-Flipping with Re-Decoding
复制标题
通过带有重新解码的噪声梯度下降位翻转来解码 LDPC 码
DOI：
--
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
Gopalakrishnan Sundararajan
中科院分区：
文献类型：
--
作者：
Tasnuva Tithi;C. Winstead;Gopalakrishnan Sundararajan
在本文中，我们考虑了在重新编码失败帧的情况下的嘈杂梯度下降位比位（NGDBF）算法的性能。 NGDBF是一种使用非确定性梯度下降搜索来解码低密度平价检查（LDPC）代码的算法。提出的重新定位程序获得了改善的性能，因为扰动在每个重新编码阶段都是独立的，因此增加了成功解码的可能性。我们检查了从IEEE 802.3AN标准重新编码为LDPC代码重新编码的好处，并发现只需要一小部分重新编码的框架才能获得显着的性能益处。使用重新编码时，NGDBF性能非常接近802.3AN代码的基准偏移量Min-sum解码器。
In this paper, we consider the performance of the Noisy Gradient Descent Bit Flipping (NGDBF) algorithm under re-decoding of failed frames. NGDBF is a recent algorithm that uses a non-deterministic gradient descent search to decode low-density parity check (LDPC) codes. The proposed re-decode procedure obtains improved performance because the perturbations are independent at each re-decoding phase, therefore increasing the likelihood of successful decoding. We examine the benefits of re-decoding for an LDPC code from the IEEE 802.3an standard, and find that only a small fraction of re-decoded frames are needed to obtain significant performance benefits. When re-decoding is used, the NGDBF performance is very close to a benchmark offset min-sum decoder for the 802.3an code.