【作者】 张小龙；

【导师】 刘德明； 张敏明；

【作者基本信息】 华中科技大学 ， 光电信息工程， 2013， 博士

【摘要】 随着高清多媒体、网络互动游戏、云接入技术等多种高带宽业务的普及应用,用户对于数据通信速率的要求越来越高,高速接入网成为了当前通信技术的研究重点。而有线电视同轴电缆网络由于覆盖面广,用户数多,因而得到了广泛的关注,并且在国内外兴起了一大批同轴接入技术,如DoCSIS (Data over Cable Service Interface Specification,有线传输数据业务接口规范)、EoC (Ethernet over COAX,以太数据通过同轴电缆传输)、EPoC (EPON Protocal over Coaxial Distribution Network,在同轴分配网络中传输EPON协议)等,同轴接入方式在三网融合的背景下已经成为了接入技术的研究重点。本论文旨在研究同轴电缆分配网的信道特性及故障诊断。在基于常见的同轴分配网现场测量结果的基础上,结合实验室的有关实验,在5MHz-1.2GHz的频率范围内,对同轴分配网信道的主要特性包括传输特性、噪声特性以及故障诊断和定位等进行了初步研究,建立了相应的信道建模,构造了新的故障诊断方法,对进一步了解同轴接入技术提供了一定的参考价值。主要内容包括：(1)对分支分配器进行了电磁建模。建模方法包括三种：基于ABCD矩阵的器件建模、基于ADS软件的器件建模、基于S参数矩阵的器件建模。对二分配器和一分支器这两个基本元件进行了三种方法的器件建模。借助于级联关系和这两个基本元件的模型,可计算出其它种类的分支分配器的器件特性。对每种模型都给出了实例,并比较了这三种模型的优缺点。基于遗传算法,对分支分配器中最核心的器件——磁芯进行了德拜级数建模,磁芯特性用于分支分配器模型中,使得器件特性的仿真结果与实测结果十分逼近。(2)对同轴电缆信道进行了多种方法的建模(曲线拟合、ABCD矩阵、S参数、ADS软件建模),提出了更加适用的基于S参数的多径信道建模方法。其模型的S参数是通过网络分析仪实测获得的,使仿真的结果跟实测的结果十分逼近,幅频曲线和相频相对误差控制在5%以内。完成了基于S参数的多径信道建模的仿真软件,在已知网络当中各器件特性的情况下,利用该仿真软件能估算出任意给定拓扑端口的传输特性,包括幅频、相频曲线、群时延和时域反射等。(3)对同轴电缆信道的噪声特性进行了分类和建模。根据噪声特性的不同,可以把影响同轴网络的噪声归纳为三种：有色背景噪声、窄带噪声和脉冲噪声。分析了各种噪声的概率密度函数,并对实测的噪声进行噪声分解；对每种噪声依据其概率密度函数进行了参数拟合,建立的模型可仿真出各种噪声的幅度谱曲线以及合噪声的分布情况。(4)提出了二种用于同轴电缆分配网的故障诊断和定位的方法。第一种方法为采用BP神经网络并结合主成分分析法(PCA),从测量的TDR曲线中分析出故障的类型(短路、断路、软故障)和故障的位置；另一种方法为借助支持向量机(SVM)进行故障诊断和定位。更多还原

【Abstract】 With the popularity of high-definition multimedia, interactive network games, cloud access technology and other high-bandwidth services applications, users increasingly high demand for data communication rate, so that high-speed access networks have become the focus of the current communication technology. Coaxial cable network with wide coverage, numbers of users, has got a lot of attention. A number of coaxial access technologies, such as DoCSIS (Data over Cable Service Interface Specification), EoC (Ethernet over COAX), EPoC (EPON Protocal over Coaxial Distribution Network), has arisen both at home and abroad. The coaxial cable access has become the focus of the access technology at the context of triple playThe thesis aims to study the channel characteristics of the coaxial cable distribution network and fault diagnosis. Based on the site test of coaxial cable networks and the experiment, the channel characteristics of the distribution network, such as the transmission characteristics, noise and the fault diagnosis and location are researched in the thesis, the frequency range is from5MHz～1.2GHz. The work is helpful for the understanding of the coaxial cable access technology. The main contents including:(1) The electromagnetic modeling of the splitter and tap is studied. There are three methods:ABCD matrix modeling, ADS software-based modeling and the S-parameter matrix modeling. These three methods are used to do model for the two-way splitter and one-way tap which the basic elements of the other splitter and tap. With the aid of the cascade relationship and the characteristics of the two basic elements, the other splitter and tap can be calculated. The example of each method is shown and the comparison is making among the three methods. Based on genetic algorithms, the Debye series modeling of the core which is the important part of the splitter and tap is studied, with the accurate modeling of the core, the simulated results are very agreement with the measured results of the transmission characteristics of splitter and tap.(2) Several methods are used to do modeling for the coaxial cable network, including curve fitting, ABCD matrix, S-parameter matrix and ADS software-based method, a novel and more comfortable method named "S-parameter based multipath channel modeling" is introduced. The S-parameter of the proposed modeling is obtained by the actual measurement with VNA, so that the simulated results are very close to the testing results, the relative error of amplitude-frequency curve and phase-frequency curve is almost less than5%. The simulated software based on the novel channel modeling is completed, which can estimate the transmission characteristic of channel between any two ports in the analytical network, including the amplitude-frequency curve, phase-frequency curve, group delay and the TDR.(3) The noise of the coaxial cable network channel is classified and modeled. According to the characteristic of the noise, the main noise in the cable network can be classified into three types, including background noise, narrow-band noise and the impulse noise. The probability density function of each noise is analyzed, and the measured noise is decomposed. Each noise is modeled according to the PDF, which can simulate the amplitude-frequency curve of each noise and aggregate noise.(4) Tow novel method for the fault diagnosis and location of coaxial cable network is proposed. The first one is based on the use of Principal Component Analysis (PCA) and BP Neural Network, which can calculate the type of fault (such as short circuit, open circuit, soft fault) and the location of fault. Another method is based on the Support Vector Machine (SVM) to diagnose and locate the fault.更多还原