A Statistical Model for Indoor SISO PLC Channels

This paper proposes a statistical channel model for indoor single-input single-output (SISO) power line communications (PLC) in the 2–80 MHz frequency band. The model follows a purely top-down strategy, since it makes no physical assumptions about the underlying power network. The frequency response is modeled as a multivariate random variable (RV) whose parameters are derived from the statistics of a set of 458 channels measured in different European countries. To this end, we firstly assess the log-normality of the amplitude response and discuss the difficulties associated to the estimation of the covariance matrix of the amplitude response at different frequencies, highlighting the need for using a regularization method. The parameters of the model are then derived by approximating the statistics of the measured channels by means of novel analytical expressions. Finally, the validity of the proposal is evaluated by comparing the average channel gain, the delay spread and the coherence bandwidth of the channels generated according to it with the ones of the measured channels and of channels generated using another top-down statistical model proposed in the literature.