Visible to the public Call for Participation: ESWeek Tutorial "Parameter-Invariant Monitor Design for Cyber Physical Systems"

No replies
Anonymous's picture


Parameter-Invariant Monitor Design for Cyber Physical Systems
ESWEEK tutorial: October 4th, 1:30 - 5:30 pm

With recent advances in low-power low-cost communication, sensing, and actuation technologies, Cyber Physical Systems (CPS) have revolutionized automated medical diagnostics and care, building energy management, and smart grids. With this revolution, dawns a new era of CPS monitoring where fusing measurements from multiple devices provides unprecedented early detection of critical events. However, some applications (e.g. medical diagnostics) explicit models and/or rich training data relating available measurements to events are unavailable or impractical. For these troublesome scenarios, this tutorial presents a parameter-invariant approach to monitor design that has been successful in developing monitors for medical conditions, building control failures, and network disturbances. Owing its mathematical origin to the robust radar signal processing literature, the parameter-invariant approach consists of three components: (1) foundations of parameter-invariant design, (2) modeling physical process for monitoring, and (3) constant false alarm rate (CFAR) hypothesis testing. To illustrate each component, the tutorial makes extensive use of case study monitors related to medical alarms (e.g. hypoxia, hypovolemia, and hypoglycemia), building heating ventilation and air conditioning (HVAC), and power grids.

  1. The foundations of parameter-invariant design consists of a design philosophy aimed at providing a monitor robust to nuisance artifacts/parameters in the data. This ultimately requires the co-design of physical models and test statistics such that maximal invariance is achieved with respect to the nuisances. In this component, we first introduce the high-level mathematical foundations for parameter-invariant design, then provide examples of three parameter-induced transformations and their respective invariant tests common in real-world monitoring applications, namely: bias, scale, and rotation.
  2. While there are many approaches to process modeling, the parameter-invariant design utilizes a lumped-parameter model derived from a first-principles model of the physical system to be monitored. The lumped-parameter model is constructed to capture the general trends associated with the monitoring event and may not accurately represent the underlying dynamics. Examples illustrating how to identify these trends will be discussed through the case studies.
  3. Through model manipulation and extensive use of null space projections and ratio distributions, a CFAR test for the critical event is designed providing near-constant performance across the population. This is achievable by first generating a statistic constrained to a class of parameter-invariant statistics, then designing a test that simultaneously monitors the event while ensuring the model accuracy and testing power is sufficient. Consistent with the model development, all concepts related to CFAR testing are demonstrated (and evaluated) through the case studies.

Intended audience:

Novice participants with an undergraduate-level understanding of linear algebra will be introduced to a new and powerful monitor design technique for CPS. Those familiar with signal processing will enjoy the elegance and rigor of the CFAR test design and gain invaluable insight into application dependent modeling for medicine, buildings, and power grids. Complementary, those with practical experience will gain insight into how high-fidelity models can be reduced to useful models for the purposes of CPS monitoring.


James Weimer is currently a Postdoctoral Researcher in the Department of Information and Computer Science at the University of Pennsylvania and holds a Ph.D. in Electrical and Computer Engineering from Carnegie Mellon University. His research interests focus on the design, analysis, and implementation of cyber-physical systems with application to medical devices and monitors, automotive systems, building energy management, and smart grids.

Oleg Sokolsky is Research Associate Professor of Computer and Information Science at the University of Pennsylvania, where he has occupied various research-track positions since 1998. His research interests lie in the application of rigorous development methods to medical cyber-physical systems, including formal modeling and verification, runtime monitoring, and safety assurance. He holds a Ph.D. degree in Computer Science from Stony Brook University.

Insup Lee is the Cecilia Fitler Moore Professor of Computer and Information Science and Director of PRECISE Center at the University of Pennsylvania. He holds a secondary appointment in the Department of Electrical and Systems Engineering. His research interests include cyber-physical systems, real-time and embedded systems, runtime assurance and verification, trust management, and high-confidence medical systems. He received a Ph.D. in Computer Science from the University of Wisconsin, Madison. He is IEEE fellow. He received IEEE TC-RTS Outstanding Technical Achievement and Leadership Award in 2008.