Study Of Seperation Tecnology For Video Signals Based On ICA In MGDM Multimode Fiber Communication System

As our country vigorously promotes the integration of three networks andpeople’s demands to data are bigger and bigger, the FTTH has begun used in a largescale. And in short optical fiber communication networks, multimode fibers haveseveral advantages like convenient installation and low cost. If we can expand thebandwidth of multimode fibers, multimode fibers will have vast prospects in the shortdistance communication networks. At this time, MGDM (Mode Group DiversityMultiplexing) technology is proposed. Through the use of many modes in multimodefibers, MGDM realizes the transmission at the same time of a variety of data inmultimode fibers, and it expands the transmission bandwidth of multimode fibers,makes it possible for multimode fiber transmission system used in the short distancecommunication networks.But the channel crosstalk in MGDM system influence the accuracy of receivedsignals, limiting the improvement of the system’s performance. Therefore, how tosolve the receiver signals crosstalk problem, become the key problem of MGDMsystem. Based on the former research, this paper used the Independent ComponentAnalysis (ICA) method to separate the video image signals at the receiver, to provethe feasibility of MGDM system in video transmission business. The main content ofthis paper is to realize the transmission of two video signals in a2x2MGDM system,and use several kinds of ICA algorithm to separate video images at the receiver, withtesting the effectiveness of the ICA algorithm. At the same time, we studied the NearField Pattern (NFP) of mode selective excitation technology at the output ofmultimode fiber, proved the correctness of the MGDM system. Paper detailed workcontents are as follows:First of all, this paper introduces the MGDM relevant technologies and systemmodels. Various reuse technologies in multimode fibers are classified arrangement,and are explained simply. The main principle of mode selective excitationtechnology which is the key technology of MGDM is introduced, and the division ofmode groups in multimode fiber is analyzed. Experiment system for testing the NFPof multimode fiber output is established, the NFPs of mode selective excitation and2x2MGDM system are tested respectively, and the correctness of the MGDM systemis proved with the ideal results. Secondly, this paper introduces the signal separation methods used in MGDMsystem. The signal separation methods in MGDM system are classified andsummarized, and obtained three categories of signal separation methods with theadvantages and disadvantages of each method. The typical method of blind signalprocessing is independent component analysis (ICA), whose basic principle, severalindependence criterions and optimization algorithms and the general processing areintroduced in detail. The ICA application in image processing is discussed, includingthe method we used in the separation of video image signals in this paper. Severaltypical types of ICA algorithm are introduced with the basic principle of eachalgorithm.Thirdly, theory simulations are studied for ICA application in video imageseparation. As for the feasibility of ICA used in image separation, this article givessome reasonable proof from the constraints conditions of the ICA. Then, mix twoimages with known matrix, and then used different ICA algorithms to separate theimages. Three ICA separation results of gray images, color images and DIY colorimages are studied, and confirmed to the effectiveness of ICA for image separation.At the same time, the effects of three ICA algorithms used in image separation arecompared.Finally, we design and implement the2x2MGDM system, achieve the videoimage transmission experiment, and use different ICA algorithms for signalseparation research to the video signals received. The multiplexing anddemultiplexing modules are optimized, and the basic models of multiplexingcomponent and demultiplexing component are designed and used in the MGDMsystem. By using two different video images for signal transmission experiment, weprove the effectiveness of the system. Different ICA algorithms are used for signalseparation to the actual received signals, and the conclusion that EGLD algorithm isthe most efficiency for actual image signals separation. In order to even more clearlydescribe the separation level of each algorithm for video image signals, this paperputs forward the concept of mean pixels, and apply it to the effects evaluation ofeach algorithm.On the basis of formers, this paper realize the video image signals transmittedin MGDM system and the use of ICA methods for signal separation, prove thecorrectness of the MGDM system and the feasibility of multi-business transmission.The results not only prove that the efficiency of ICA application in MGDM systemfor signal separation, but also has important meaning for the process of commercial applications of MGDM system.