Research On Noise Modeling And Elimination Of Multi-input Multi-output Power Line Communication

Benefit from the wide distribution of power lines,no need for rewiring and economical convenience,power line communication(PLC)has received a wide range of popular research and applications in smart grids and home networks.With the development of traditional PLC technology and the continuous improvement of communication requirements for PLC,the researchers extends the traditional single-input single-output power line communication(SISO-PLC)to multi-input and multi-output power line communication(MIMO-PLC).In order to reduce the noise interference of MIMO-PLC,it is necessary to study the modeling of MIMO-PLC channel noise first.At present,there are two main methods to model PLC channel noise,one is to model different kinds of noise in PLC channel separately,the other is to model PLC channel noise as a whole,but there are some problems,such as heavy workload,cumbersome design and error accumulation.In order to improve the shortcomings of the existing methods,this thesis is based on the correlation between the MIMO-PLC channel noise,and the support vector regression(SVR),the least Squares support vector machine(LS-SVM),wavelet neural network(WNN)and length-time memory network(LSTM)is applied to the modeling research of MIMO-PLC channel noise.Firstly,the principle of these four models is introduced,secondly,the content of data processing,parameter selection and model structure are analyzed,and finally the results of four models are evaluated by theR~2-score,mean absolute error(MAE)and average absolute percentage error(MAPE).According to the evaluation index,four models are compared and analyzed.finding that all of the methods can model the noise in the MIMO-PLC channel effectively.In the aspect of weakening noise interference,the existing methods need to design filters or filter groups for different types of noise respectively,making the receiving end of PLC communication system complicated.In order to improve this problem,in this thesis,adaptive noise cancellation method is used to reduce the interference of noise in MIMO-PLC on the basis of noise modeling.In order to verify the validity of this method,a complete MIMO-PLC communication system with designed receiving end and sending end is constructed,in which the transmitter adopts convolution coding,perturbation,interleaving and orthogonal frequency division multiplexing(OFDM)technology,the receiving end adopts the corresponding decoding technology and adaptive Noise elimination technology.The simulation results show that the adaptive noise cancellation method can effectively reduce the bit error rate of the MIMO-PLC system,and the adaptive noise cancellation method can effectively reduce the bit error rate of MIMO-PLC system up to 15 dB,and improve the communication quality of the system.The superiority of the noise modeling method proposed in this thesis for MIMO-PLC noise modeling is also verified.