Impulsive Noise Mitigation With Interleaving Based on MUSIC in Power Line Communication

In power line communication, impulsive noise (IN) greatly degrades the performance of signal transmission. Generally, an IN model adopts Middleton Class A or Bernoulli Gaussian model with the assumption that IN is sparse and memoryless. However, IN collected in fields appears in bursts. In this paper, interleaving in time domain is applied to create a special structure in frequency domain for IN burst. Based on the special structure, IN support is determined by using multiple signal classification (MUSIC) and its amplitude is estimated by least squares method in frequency domain, exploiting the noise information on null subcarriers in orthogonal frequency division multiplexing system. In addition, sparse Bayesian Learning algorithm with IN support based on MUSIC is also investigated. Simulations are conducted to evaluate the proposed IN estimation performance, with varying ratio of IN power to background noise power, and number of null subcarriers.