Convolutional Coded Bayesian Inference Based Channel Estimation in Power Line Communication Systems

Power line communication (PLC) channel as a medium for high speed data communication transmission based on orthogonal frequency division multiplexing (OFDM) is considered. It is an environment with frequency selective and multipath fading features which has been contaminated by impulsive noise. These deficiencies in power line communications degrade the accuracy of channel estimation. In this article, an efficient channel estimation method based on bayesian inference is presented. A new proposed kernel function with proper hyper-parameters in relevance vector machine (RVM) is used to estimate the PLC channel impulse response. Bit error rate for hard and soft decisions in Viterbi decoding corresponded to convolutional coded data and mean square error (MSE) are evaluated and compared. Proposed channel estimation algorithm achieves good results respected to recently reported approaches as Huang channel estimation method. It is shown that about 8dB improvement in MSE with respect to Huang method is achieved. Also, for bit error rate (BER=10), about 2.2dB and 1.8dB enhancements in signal to impulsive and background noises ratio (SNIR) respected to Huang method with soft and hard decisions are obtained, respectively. Keywords—Convolutional coding, impulsive noise, multipath effects, OFDM, power line communication, RVM.