Maximum-likelihood channel estimation in presence of impulsive noise for PLC systems

Abstract

Impulsive noise in power lines usually bears higher power spectral density than the background noise. Even a sample of such noise, if not mitigated, destroys symbols in all the subcarriers affecting channel estimation and information detection. In this paper, we study two maximum-likelihood approaches to estimate the channel in the presence of impulsive noise for an OFDM based PLC system. In both approaches, we approximate the impulsive noise first and use it to determine the channel frequency response. In the first approach, we assume the estimated noise to be random and perform the channel estimation assuming that its statistics are given. In the second approach, the impulsive noise is assumed to be deterministic and the channel frequency response is determined after proper subtraction from the observations. As will be shown by simulation results, the estimator that assumes noise to be random generally performs better than the one that treats the estimated impulsive noise as a deterministic quantity.

On channel estimation for power line communication systems in the presence of impulsive noise

“…In this section we analyze and compare the performance of both estimators in varied scenarios by exploiting the derived covariances in Eqs. (18) and (27) . To begin with, we first consider an ideal scenario where the support of the IN is perfectly estimated and characterize the variance of both channel estimators.…”

“…In order to precisely estimate the channel in the harsh power line environment where the IN is prominent, a better performance than the conventional channel estimation strategies can be achieved, if instead of trying to suppress the IN affected sam ples in the received signal, the IN is estimated and used in the process of the estimation of the channel [18] . To do so, the sparse IN present in the received signal can be estimated by using the signal processing schemes like the ones proposed in [19][20][21] can be used.…”

“…Maximum-likelihood (ML) estimation of the channel by exploiting the estimated IN for PLC systems has not yet been widely explored. In one of our prior works, reported in [18] , we showed some preliminary results on how the estimated IN samples can be used to estimate the channel for PLC. The results show that channel can be estimated with better precision in the presence of IN by exploiting the ML technique.…”