Parametric and nonparametric methods for power line network topology inference

Frequency domain reflectometry (FDR) has recently been proposed as a method for inferring the topology of power line communications (PLC) networks. In this paper, we introduce and compare two approaches to process FDR measurements for topology inference. The first approach is based on peak-detection using the inverse Fourier transform of FDR measurements. We refer to this method as the nonparametric method. The second approach is based on a parametric model, which under reasonable assumptions makes Prony-type methods applicable to the problem at hand. The advantages and the limitations of both methods are determined and discussed. Simulation results demonstrate the performance of each method.