基于CNN-LSTM的光通信信道自适应均衡算法研究  

作　　者：文彬 鞠理 褚君达 

机构地区：[1]92730部队 [2]91001部队

出　　处：《现代传输》2025年第2期70-74,共5页Modern Transmission

摘　　要：光纤通信以其超高带宽、低损耗的优势,已成为全球信息传输的核心技术。但是,光纤信道中固有的色散效应、非线性效应以及信号噪声等问题逐渐凸显,严重影响了信号的传输质量。为了有效补偿这些信道损伤,提高光通信系统的传输性能,信道均衡技术成为研究的关键问题之一。构建基于CNN-LSTM的光通信信道均衡模型,以提高复杂光纤信道条件下的均衡性能。实验结果表明,CNN-LSTM均衡相比CMA均衡误码率降低了约79%,相比DFE均衡误码率降低约63%,相比CNN均衡误码率降低了44%,同时SNR增益较传统均衡方法提升了2~4d B。此外,从均衡后的星座图来看,CNN-LSTM均衡后的符号点更加集中,表明其能够更准确地恢复原始信号,提高系统的传输质量。Fiber optic communication,with its advantages of ultra-high bandwidth and low loss,has become the core technology for global information transmission.However,the inherent dispersion effect,nonlinear effect,and signal noise in fiber optic channels have gradually become prominent,seriously affecting the transmission quality of signals.In order to effectively compensate for these channel damages and improve the transmission performance of optical communication systems,channel equalization technology has become one of the key research issues.Construct an optical communication channel equalization model based on CNN-LSTM to improve the equalization performance under complex fiber channel conditions.The experimental results show that CNN-LSTM equalization reduces the bit error rate by about 79%compared to CMA equalization,by about 63%compared to DFE equalization,and by 44%compared to CNN equalization.At the same time,the SNR gain is improved by 2-4 dB compared to traditional equalization methods.In addition,from the balanced constellation diagram,the CNN-LSTM has more concentrated symbol points after equalization,indicating that it can more accurately recover the original signal and improve the transmission quality of the system.

关 键 词：光通信 信道自适应均衡算法 CNN-LSTM模型 

分 类 号：TN929.1[电子电信—通信与信息系统]