| Mode group diversity multiplexing (MGDM) aims at creating independent, parallel communication channels over a single GI-MMF using several groups of modes and can greatly improve the capacity of short-reach MMF networks. It is a key technology to reduce the cost of optical access network, upgrade the existing access network, and achieve the FTTB as well as the FTTH. The proposal of MGDM communication technology based on the GI-MMF can effectively resolve the bandwidth limitation of MMF, upgrade rapidly and economically on the basis of existing multimode fiber and play a vital role in the current development of the optical fiber network. The most important problem to be solved is the crosstalk between the channels which can be called de-multiplexing, which has a great impact on the bandwidth of the MGDM system. How to implement simple and effective de-multiplexing while maintaining the low-cost of the system is the main task in this thesis.This paper has done a lot of work on the further development of this technology on the basis of previous research. With the starting point of the low-cost, practical and easy to use, a system model of MGDM is built, and a 2×2 MGDM experimental system based on 1km-long silica GI-MMF with core–cladding diameter 62.5/125μm is designed and realized, which achieve communications successfully while focusing on the alleviation of crosstalk between channels. It is the first time to propose the independent component analysis (ICA) as the de-multiplexing method in MGDM channels, of which FastICA algorithm based on the transient linear mixed model is adopted. According to the characteristics of the MGDM system, channel identification codes are used as an innovative approach to overcome the major defect of ICA, uncertainties of sequences and symbols of signals, which can significantly reduce complexity while guaranteeing the de-multiplexing performance. The works in this paper are as follows:Firstly,a mathematical model of the IM-DD M×N MGDM system based on GI-MMF is developed, and the relationship between N input signals and M output signals can be described with a real-valued M×N channel matrix H(t). Besides, the expression of the electric field plane wave of a single mode is derived from Maxwell's equations, and the theory of mode groups is introduced. The optimal excitation model and spatial selective detection model of MGDM system are also developed. The paper analyzes the frequency response of MGDM system as well, and comes to the conclusion that the elements of channel impulse response and frequency response matrix are real under the proposed model. The impaction of system noise is studied as well as the power loss it caused. MGDM system bandwidth is investigated and M=N is the best conclusion.Secondly, the selective mode excitation principle is analyzed, quantified the concept of selective mode excitation and mode groups in allusion to idiographic GI-MMF. Moreover, the differences between NFPs of GI-MMF mode groups are described from the simulation in electromagnetic theory and light theory. Then, mode group multiplexer and de-multiplexer are produced for the 2×2 IM-DD MGDM experimental system, and are optimized through experiments. Therefore, it is necessary for a practical MGDM system to have the electrical signal processing. The mode conversion phenomenon of GI-MMF'fusion point is analyzed.Thirdly, the basic conception of ICA and the estimation principle are introduced in detail, and then the feasibility of the de-multiplexing of MGDM system with ICA is analyzed. In order to overcome the two major defects of ICA as it was mentioned above, an innovation in this work that channel identification codes are used considering the characteristics of the MGDM system is given, which can greatly reduce complexity of the system while guaranteeing the de-multiplexing performance of system. The corresponding FastICA algorithm based on negentropy is developed.Finally, a 2×2 IM-DD MGDM experimental system is built and is used to transmit different rates of digital signals and digital video signal directly which showed the limitation of spatial selective detection. Under the condition of off-line, de-multiplexing of the real data which are gathered from the MGDM experimental system are realized with the improved FastICA algorithm, and the performance of this method is similar to that of the zeros forcing method. The performance of the combination of this algorithm with the Butterworth low-pass filter is studied, and the system can work at 5dB (SNR) with the implementation of this de-multiplexing method.In the short-distance high-speed communications network, multimode fiber communication system which based on the MGDM has unique advantages of fiber-optic network and the same function as WDM which the SMF are used, which can considerably enhance the system capacity and solve the bandwidth limitation of multimode fiber effectively. However, the requirements in the system are less critical than WDM on the light source, and the installation and maintenance of multimode fiber are more convenient and its components are reasonable in price. The adoption of the proposed method can make the de-multiplexing of MGDM system easier only with the information of received signals, and avoid the complexity of matrix inversion without using pilot signal and estimating the channel matrix, so the system complexity may be lower and so as the cost. |
PDF Full Text Request