Research On Channel And Noise Modeling For Low-Voltage Broadband Power Lines

Low-voltage power line communication(PLC)technology using existed power line as communication media has been payed great attention worldwide because of its wide coverage,convenient networking without requiring installation of new communication equipments,etc.PLC technology is widely used in people's daily life,such as smart meter reading,home automation,remote street lighting control ans so on,it is also a key communication technology for data transmission in smart grids.However,PLC is a unique communication system which uses power line to transmit high-frequency signals,which is very different from traditional communication systems because of the PLC structure and transmission properites,such as low-pass characteristic,noise interference,electromagnetic compatibility and time-variation.Therefore,to realize high-speed and reliable PLC is a challenging task.In order to use various physical layer technologies in PLC to improve its transmission quality and system performance,simple and accurate PLC channel and noise models are required by fully understanding the PLC characteristics.However,existed PLC channel and noise models have many disadvantages,which are required to be optimized and improved.The main research contents and contributions of the work are summarized as follows(1)This paper proposes a novel low-voltage PLC channel model using graph theory.The first K shortest paths of PLC channel are solved by recursive operation of Iterative Enumeration Algorithm(IEA)so that the transmission function of power line channel can be simulated simply and effectively.The modeling method is compared with two traditional PLC channel modeling methods by transmission line(TL)theory and parameter matching based.(2)This paper proposes a low-voltage broadband MIMO(Multi-Input Muiti-Output)PLC channel modeling method,which is extended by SISO(Single-Input Single-Output)PLC channel models based on graph theory.The modeling efficiency is verified by measurement data and compared with two traditional MIMO PLC channel models.(3)Aiming at the shortcomings of existed PLC noise models,machine learning is applied in modeling,and we propose two low-voltage PLC noise models based on wavelet neural network and least squares support vector machine(LS-SVM),respectively.The effectiveness of the two noise models is validated by measurements and applied in building new and simple noise generators.(4)Considering the complicated PLC network structures and varieties,the influence of power-line network structures,such as power line length,number of node,branch length,branch load and so on is studied by frequency-and time-domain channel responses.Furthermore,the PLC channel statistic properites such as root mean square delay spread(DS-RMS),coherence bandwidth(CB)and channel capacity etc.are investigated with respect to the power-line network structures.(5)To increase PLC transmission rate and coverage,a multi-hop broadband relay-aided PLC system is studied in the thesis.Locations of the installed relay nodes are derived for the actual power-line networks and also validated by simulations.(6)By maximizing the signal to noise ratio(SNR),the best matched impedances of the transceiver are derived,which make the PLC with better system performance.