Article Source

• The Power of Graph Signal Processing

The Power of Graph Signal Processing

Abstract

The theory of graph signal processing (GSP) was formulated to extend fundamental insights that come from the frequency analysis for time series to the domain of signals indexed by graphs. The power grid is one of the foremost examples of a large-scale man-made network. It is therefore natural to see measurements from the power grid as graph signals and model power grid measurements using tools from GSP. In this talk, the paradigm of GSP will be discussed in the context of the power grid where a spatio-temporal generative model for voltage measurements is developed from first principles by leveraging system-level knowledge of power systems. How GSP based tools can be effectively used for different purposes in the power grid is demonstrated using both synthetic data and real-world measurements. Ultimately, the GSP based modelling strategy for the power grid illustrates how a variety of data such as those coming from financial or social networks can be analyzed using similar tools presented in this talk.