Inference and communications over power lines

Power line communications (PLC) refers to the technology that reuses power supply cables for communications. Its origins go back to the early 1900's, and it has long been used for control and automation by electricity utilities. PLC has attracted considerable attention for supporting smart grid applications. Since it reuses the existing grid infrastructure, it offers cost advantages over alternative communications methods and gives electric utilities control over the communications medium. Furthermore, the "through-the-grid" property of PLC extends its possible use beyond mere communications. Since the PLC signals are bound to travel through the power grid, they can also be used for inference tasks, such as online diagnostics of power line integrity. In this thesis, we study two different problems that are related to PLC. The first problem is inference of power line topology. That is how to reconstruct the power network infrastructure using certain measurements. For this problem, the "through-the-grid" property of PLC is exploited. We investigate two solutions for this problem based on single-ended or multiple-ended measurements. We propose algorithms for both cases. Simulation results demonstrate the successful and accurate reconstruction of the grid topology by the proposed algorithms. The second problem is crosstalk elimination for PLC. For this problem, experiments were performed to measure the crosstalk and compare the results with analytical models. The results from the measurement campaign were used to evaluate the possible gains by the use of multiuser detection methods for crosstalk cancellation. Measurements and simulation results show the effectiveness of the proposed algorithms and methods.Applied Science, Faculty ofElectrical and Computer Engineering, Department ofGraduat