Refined Density Evolution Analysis of LDPC Codes for Successive Interference Cancellation
Refined Density Evolution Analysis of LDPC Codes for Successive Interference Cancellation
复制标题
DOI：
10.1109/globecom46510.2021.9685529
复制
发表时间：
2021-12
期刊：
影响因子：
--
通讯作者：
Satoshi Takabe;T. Wadayama;Masahito Hayashi
中科院分区：
文献类型：
--
作者：
Satoshi Takabe;T. Wadayama;Masahito Hayashi
连续干扰消除 (SIC) 是高斯多址信道 (GMAC) 的基本解码技术。在SIC中，每个用户的发送信号被单独解码。在本文中，我们分析了 $N$ 用户 GMAC 上实际低密度奇偶校验 (LDPC) 码的 SIC 解码的渐近解码阈值。传统上，基于所谓的信道近似(CA)来评估解码阈值，其中SIC解码器的每个解码级的信道模型通过简单的高斯噪声信道来近似，从而导致解码阈值的误差。为了避免这种情况，我们提出了一种称为 DE-SIC 的精细密度演化 (DE) 分析，它使用与每个解码阶段相对应的混合高斯噪声通道。我们通过解决 GMAC 中的接收功率优化问题并将其与使用 CA 的传统 DE 分析进行比较来演示 DE-SIC。结果表明，DE-SIC 准确地评估了解码阈值，而传统分析则低估了阈值。
Successive interference cancellation (SIC) is a fundamental decoding technique for Gaussian multiple access channels (GMAC). In SIC, transmit signals of each user are separately decoded. In this paper, we analyze an asymptotic decoding threshold of SIC decoding for practical low-density parity-check (LDPC) codes over $N$-user GMAC. Conventionally, the decoding thresholds are evaluated based on the so-called channel approximation (CA) in which the channel model of each decoding stage of a SIC decoder is approximated by simple Gaussian noise channels resulting in errors of decoding thresholds. To avoid this, we propose a refined density evolution (DE) analysis called DE-SIC which uses mixed-Gaussian noise channels corresponding to each decoding stage. We demonstrate DE-SIC by solving a received power optimization problem in GMAC and comparing it to the conventional DE analysis with CA. The results show that DE-SIC accurately evaluates the decoding thresholds whereas the conventional analysis underestimates the thresholds.