| Title | : | Anomaly detection in data-driven Cyber-Physical Systems |
| Speaker | : | Srinivasan Iyengar (University of Massachusetts Amherst) |
| Details | : | Wed, 1 Aug, 2018 3:00 PM @ A M Turing Hall |
| Abstract: | : | Technological advances have improved our ability to deploy pervasive sensing and actuation in our physical world. This proliferation has resulted in the emergence of cyber-physical systems (CPS) where cyber systems (computing, sensing) interact with the physical world and humans in unique and exciting ways. CPS, through sensing and embedded computing, are employed in new application domains such as smart grids, wearables, precision agriculture, and smart transportation to generate rich and vast amounts of data. The design of such data-driven systems leads to several research problems.
In this talk, I will focus on the problem of anomaly detection in the data-driven CPS. Initially, I will explain the fundamental limitations of applying traditional anomaly detection used in systems (e.g., networking, cloud computing) to the CPS context. I will present an alternative approach that overcomes these limitations and is amenable to the dynamism and scale of CPS. I will primarily describe two distinct anomaly detection problems from the energy domain that leverage this approach. First, I will present SolarClique, an automated technique to identify anomalous power generation from solar installations. SolarClique enables large-scale anomaly detection of millions of solar sites without the need for installing additional sensors. Second, I will introduce WattHome, a model-based method to detect inefficiencies in buildings using their energy consumption data. Further, I will present the results of a city-scale evaluation performed in collaboration with a local utility, where WattHome successfully identified causes of energy inefficiency of several thousand buildings. Both these applications highlight cases where scalable approaches can be utilized to build CPS-enabled sustainable societies. I will end with my future research directions in the field of cyber-physical systems and how it can create significant opportunities for interdisciplinary work. Speaker’s Bio: Srinivasan Iyengar is a Ph.D. candidate in the College of Information and Computer Sciences at the University of Massachusetts Amherst, where he is advised by Prof. Prashant Shenoy. His research interests span several areas of computer systems, including cyber-physical systems, distributed systems, and embedded systems. His current research focuses on designing data-driven cyber-physical systems, with a particular emphasis in domains such as smart cities, smart buildings, and transportation. |
