Control-System Interactions in Cyber-Physical Infrastructures
Sandip Roy — Washington State University
Wednesday, January 16, 2019
ABSTRACT: Cyber-technologies are enabling a new paradigm for control in modern infrastructure networks, centered around client-catered and mission-adaptive decision-making. While this new paradigm holds remarkable promise, it also brings forth fundamental new challenges in infrastructure controls engineering, including growing democratization of control authority, increasing vulnerability to cyber attacks and failures, dependence on ad hoc sensing, and concern about system-wide cascading events. In this talk, the new paradigm and attendant challenges in infrastructure-network control are illustrated in two very different application domains: deployment of new wide-area controls to damp oscillations in the bulk power grid, and management of emergent antibiotic-resistant emergent infections at multiple scales. Then, a research program for assessing and designing infrastructure controls is envisioned, which is based on understanding interactions among control systems in dynamical networks. Specifically, four directions of work are overviewed: 1) input-output (channel) analysis for dynamical networks, 2) control-channel interaction assessment, 3) channel-preserving model reduction, and 4) ad hoc sensing-based control. Preliminary theoretical results and contributions to several application domains, including the two motivating applications, are presented.
BIO: Sandip Roy is a Professor in the School of Electrical Engineering and Computer Science at Washington State University. His research is concerned with developing techniques for the estimation and control of network dynamics, and applying these techniques to support the wide-area management of cyber-physical infrastructures. This research has yielded algorithms and decision-support software that are being prototyped in the United States air traffic management system and the Western U.S. power grid. Recently, he has been also interested in developing network-controls techniques for epidemiological and neuroscience applications. The outcomes of the research are described in about 70 journal papers and 130 conference papers across multiple disciplines.
Hosted by: Paul Bogdan