SVD-based de-noising and parametric channel estimation for power line communication systems

Abstract

The frequency response of a power line communications (PLC) channel can be described quite accurately as a superposition of complex exponential functions. For this kind of signal there exist a number of methods based on the singular value decomposition (SVD) of a related Hankel matrix, that aim at either a de-noising or even a parameter estimation of the exponentials. We review these methods, extend one of them, and employ them for orthogonal frequency-division multiplexing (OFDM) based PLC channel estimation. Different from the more conventional channel estimation methods based on linear filters or time-domain sparsity, the methods we consider benefit from exploiting the parametric sparsity of the channel frequency response and are hence optimal in that sense. By means of simulation we compare their performance for PLC channels and discuss the suitability for different pilot models.