节点文献
少模光纤模式复用系统解复用算法研究
Research on Mode Demultiplexing Algorithm in Few Mode Fiber Mode Division Multiplexing System
【作者】 赵玲;
【导师】 胡贵军;
【作者基本信息】 吉林大学 , 通信与信息系统, 2015, 硕士
【摘要】 随着信息化时代的飞速发展,在高清视频、云文件存储、云计算、移动互联网和其他信息技术的刺激下,全球数据流量正以每年超过50%的速度增长,这对作为互联网传输骨干的长距离光纤传输网络提出了更高的容量要求。然而,随着现行WDM系统信道数目和信号调制阶数的升高,单根光纤内传输的信号功率日益增大,这加剧了单模光纤内固有的非线性效应和光纤放大器的ASE噪声,从而使单模光纤通信系统的容量逐渐达到信息论上的极限,寻找一种有效的系统扩容方式也成为光纤通信领域亟待解决的一个难题。在这种背景下,作为一种具有巨大应用潜力的系统扩容方式,模式复用技术引起了国内外研究者的高度关注。通过利用模式这一新的自由度,结合适用于长距离信号传输的少模光纤,模式复用技术可以使现有的光纤传输系统容量成倍提升。然而,由于在少模光纤中存在色度色散(CD)、模式耦合(MC)、模式群时延(MGD)等损伤,它们之间的相互影响使模式复用系统的信道状况复杂化,导致接收端的解复用难度大幅度上升,成为少模光纤模式复用技术中一个关键性的难题。因此,本文的主要研究目标是通过对少模光纤信道特性的理论分析,开发一种有效且具有更低复杂度的模式解复用DSP算法。围绕上述中心目标,本文首先简要介绍了少模光纤的概念和少模光纤中的传输模式,分析了少模光纤中的主要串扰来源——模式耦合和模式群时延对模式复用信号的具体影响,并在以上理论基础上建立少模光纤链路的数学模型。随后,分别对用于模式解复用的时域和频域LMS算法进行了研究,对比了两者的算法复杂度和解复用效果。最后,提出将频域ICA算法用于模式解复用,具体介绍了F-ICA算法的模式解复用流程。在不同信噪比和模式耦合强度条件下验证了F-ICA算法的解复用性能,并将其算法复杂度与F-LMS算法做出对比分析。相关的具体工作如下:首先,对长距离模式复用系统的传输媒质——少模光纤的模式传输特性进行了介绍和理论推导。从光波导学角度证明了光纤中模式之间的独立性,明确了模式复用的合理性和可实现性。深入研究了少模光纤中两种最重要的传输损伤:模式耦合和模式群时延对模式复用信号的影响。其次,在色度色散、模式耦合和模式群时延等因素存在的条件下,对双模光纤中的两个非简并模式(LP01和LP11模)进行了建模理论的分析。在理论基础上,通过Mode Solver计算少模光纤参数,利用VPI仿真平台,结合Matlab软件编程,构建了双模光纤传输链路的仿真模型。介绍了模式复用系统的基本原理和结构,并应用建成的双模光纤链路模型在VPI平台内搭建了22少模光纤模式复用仿真系统。再次,对模式解复用中常用的时域和频域LMS算法原理进行了介绍,分别分析了这两类算法的计算复杂度与模式复用系统中模式数目和差分模式时延长度的数学关系。利用上述两种算法进行模式解复用仿真实验,通过输出信号的星座图和眼图观察了两种算法的解复用效果,在光纤中信噪比变化的情况下对比了二者的误码性能。最后,提出将频域ICA算法用于模式解复用。详细介绍了F-ICA算法的模型和分离原理,利用所开发的F-ICA算法在仿真系统内实现模式解复用。通过解复用后信号的星座图和眼图验证了F-ICA算法的解复用效果。在耦合系数、数据长度等参数变化的情况下观察了F-ICA解复用后信号的误码率,证明了算法的稳定性。将其误码性能和算法复杂度与上一章中的F-LMS算法进行对比,结果证明了F-ICA算法在误码性能同等的情况下,具有复杂度上的优势。
【Abstract】 With the rapid development of the information age, spurred by high-definitionvideo streaming, cloud storage, cloud computing, mobile networking and otherinformation technologies, worldwide data traffic is growing at a rate estimated toexceed50%annually. It brings higher capacity requirements to long-haul optical fibertransmission network, which forms the backbone of the Internet. However, with thechannel number and signal modulation order increasing in the existing WDM system,the coupling power within one single mode fiber is scaling, and as a result, thetransmission capacity per fiber is now approaching fundamental information-theoreticlimits imposed by optical amplifier noise and by the nonlinear response of the silicafiber medium. Therefore, finding an effective way to expand the system capacity hasbecome a urgent problem to be solved in optical fiber communication field.In this context, as a kind of capacity expanding method with great potential forapplication, mode division multiplexing technology has attracted the attention ofresearchers both inland and abroad. Utilizing the spacial modes as new degree offreedom, combined with few mode fiber suitable for long distance signal transmission,mode division multiplexing technology can multiply the capacity of long-haulcommunication systems. However, due to the joint effects of chromatic dispersion,mode coupling and mode group delay, the channel condition in mode divisionmultiplexing system is far more complex, which makes mode demultiplexing one ofthe most critical problems in mode division multiplexing technology. Therefore, themain objective of this paper is to develop an effective digital signal processingalgorithm with low computational complexity to implement mode demultiplexingthrough theoretical analysis on few-mode fiber channel characteristics.This paper briefly introduces the concept of few-mode fiber, analyzes the primarycrosstalk source in few mode fiber: mode coupling and mode of group delay’sconcrete impact on mode multiplexing signals, and establishes the mathematicalmodel of few-mode fiber link based on the above theory. Then, the time-domain andfrequency-domain LMS algorithm used in mode demultiplexing are studiedrespectively, their computational complexity and demultiplexing performance arecompared as well. At last, the frequency-domain ICA algorithm is proposed asdemultiplexing algorithm in mode division multiplexing system. The principle ofF-ICA algorithm is described in detail. At the same time, the symbol error rateperformance of F-ICA is estimated under different optical signal noise ratio and mode coupling factor, and the proposed algorithm’s computational complexity is comparedwith frequency-domain LMS algorithm. The details of this paper are as follows:Firstly, the mode characters of few mode fiber, which is the transmission mediumof long distance mode division multiplexing system, are introduced. Theorthogonality of modes in few mode fiber is demonstrated from the perspective ofoptical waveguide theory to confirm the reasonableness and realizableness of modedivision multiplexing. The two primary impairments in few-mode fiber transmission:mode coupling and mode group delay are discussed and their effects on modemultiplexing signals are studied.Secondly, establish the theoretical model of the two non-degeneracy modes(LP01andLP11mode) in dual-mode fiber with the effect of chromatic dispersion, modecoupling and mode group delay. Calculate the parameters of few mode fiber withMode Solver and build the dual-mode fiber simulation model with VPI transmissionMaker and Matlab programming on the basis of few mode fiber channel modelingtheory. Introduce the basic principles and structure of mode division multiplexingsystem and construct a22few mode fiber mode division multiplexing simulationsystem on VPI platform using the built up few mode fiber transmission model.Thirdly, the principle of typical model demultiplexing algorithms: time-domainLMS algorithm and frequency-domain LMS algorithm are introduced separately, therelationship between the two algorithms’ computational complexity and mode numberor differential mode group delay is analyzed. The two algorithms are used toimplement mode demultiplexing in simulation system and their performance areobserved through the constellation and eye diagram of output signals. Compare theirsymbol error rate performance under different optical signal noise ratio.Finally, the frequency-domain ICA algorithm is proposed as mode demultiplexingalgorithm. The separation model and principle of F-ICA algorithm are described indetail, and F-ICA algorithm is successfully applied in the mode division multiplexingsimulation system. The demultiplexing performance of F-ICA is confirmed throughconstellation diagram and eye diagram. Furthermore, the symbol error rate of F-ICAalgorithm is observed in case of different coupling factors and data length, and thestability of the proposed algorithm is verified. F-ICA’s symbol error rate performanceand computational complexity is compared with F-LMS algorithm in the previouschapter. The results show that under the same symbol error rate circumstance, theF-ICAalgorithm has certain advantage of lower computational complexity.