Keynote Speaker

RAFFAELE ROMAGNOLI
Title:

Software Rejuvenation for Secure and Safe Control of Cyber-Physical Systems


Abstract:

Cyber-physical systems (CPSs) are ubiquitous in many domains, including energy delivery, intelligent transportation, and health care, thanks to recent advancements in computing, sensing, and networking technologies. One of the main challenges is to guarantee safe and secure operation of these critical CPS from possible malicious attacks and malfunctions. In this talk we present an architecture and run-time strategies for safe and secure control of CPS based on software rejuvenation. We show how to combine Lyapunov theory to preserve safety while software rejuvenation is used to guarantee security. We also deal with the liveness property of the proposed method for tracking control problems, and robustness with respect to disturbances, modeling uncertainties, and state estimation errors. The developed method is applied on a 6DOF UAV controlled by the PX4 flight controller.

Short Biography:

RAFFAELE ROMAGNOLI received the Ph.D. degree in control system and automation specializing in optimal and robust control system theory from Universit\'{a} Politecnica delle Marche (UNIVPM), Ancona, Italy, in 2015. From October 2015 to December 2017, he was a postdoctoral researcher with the Department of Control Engineering and System Analysis (SAAS), Universit\'{e} Libre de Bruxelles (ULB), Brussels, Belgium, working on the control of Li-ion batteries. Since 2018 he is a researcher with the Department of Electrical and Computer Engineering, Carnegie Mellon University, Pittsburgh, PA, USA, and is involved in several projects in collaboration with Carnegie Mellon’s Software Engineering Institute (SEI). His main research interests are related to safe and secure control of cyber-physical systems with application to autonomous vehicles.

Kalyan Vaidyanathan
Title:

Developing Optimal Software Rejuvenation Strategies based on Machine Learning Techniques


Abstract:

Over the last two decades, several well-known studies on software reliability have reported and examined in detail the phenomenon of increasing failure rate and/or progressive performance degradation in long-running software systems known as software aging. To counteract and mitigate software aging, software rejuvenation has been proposed, which involves terminating an application or a software system, cleaning its internal state and/or its environment, and restarting it. Optimal rejuvenation design depends on the application availability requirement, the application failure rate, costs of scheduled and unscheduled maintenance, and the rejuvenation performance impact. This talk will initially survey software rejuvenation optimization techniques based on supervised and unsupervised machine learning algorithms, discuss the types of data and potential challenges. The later half of the talk will focus on optimal preventive maintenance strategies that employ reinforcement learning, where applications provide feedback through a reward structure.

Short Biography:

Kalyan Vaidyanathan currently works for BAE Systems Inc. where he leads a team focussed on applied machine learning to solve problems in multiple domains including imagery, text, geospatial applications and manufacturing. He received his Ph.D. degree from Duke University in Electrical & Computer Engineering, working on developing and implementing analytic and data-centric strategies for optimal software rejuvenation. He has published highly-cited works in these areas and has several patents to his credit.