信噪比(成像)
块(置换群论)
调制(音乐)
概率逻辑
计算机科学
算法
能量(信号处理)
分解
信号(编程语言)
匹配(统计)
方案(数学)
分布(数学)
拓扑(电路)
电子工程
数学
电信
物理
工程类
统计
声学
人工智能
组合数学
生态学
数学分析
生物
程序设计语言
作者
Xuezhen Wang,Lishan Yang,Chenglin Bai,Hengying Xu,Danping Pan,Weibin Sun,Yining Zhang,Zukai Sun,Pengfei Li
标识
DOI:10.1109/acp/poem59049.2023.10368738
摘要
Benefiting from its shaping gain and flexible information rates, probabilistic shaping (PS) is on track to become a popular coded modulation standard in optical communication systems. Though producing amplitude sequences with a fixed em-pirical distribution, constant composition distribution matching (CCDM) suffers from high rate losses for small block lengths. In this paper, a multi-level decomposition enumerative sphere shaping (MDESS) scheme is proposed to implement PS for the application in the short blocklengths regime, which is a practical solution of enumerative sphere shaping (ESS) methodology and could always ensure the ultimate energy efficiency of the output sequences. Compared with traditional ESS, MDESS makes full use of low-energy sequences and small-amplitude sequences. We demonstrate the superiority of MDESS through a 28 Gbaud PS simulation system, which successfully reduces the rate loss and improves the optical signal-to-noise ratio (OSNR) gain without increasing the complexity.
科研通智能强力驱动
Strongly Powered by AbleSci AI