Estimation of Impulse Noise Parameters in Power Line Communications Channel Based on Artificial Neural Networks

Asynchronous impulse noise is among the factors that significantly degrade the performance of power line communication systems and it is very difficulty to model such noise's occurrence. In this paper, a partitioned Markov chain is employed to model the impulsive noise events in power line communication channels. The systems' states are grouped into two classes according to the occurrence of impulse events. One class represents the occurrence and the other is impulse-free. At the same time, the transition probability matrix of the partitioned Markov chain is constructed as normal. In order to determine the parameters in Markov chain, a certain neural network is used to minimize the computation load. The simulation results are also compared with the other researchers' work and a good agreement can be found which verifies the correct and good performance of the proposed method