Computation-Based Probabilistic Validation of Complex Cyber-Controlled Systems
Abstract
The project research was concerned with verification of stochastic hybrid systems with temporal logic performance specifications. The major technical theme of the research was the leveraging of the computational tools from both probabilistic physical system verification, and those of discrete system verification, to create a computationally effective and automated verification methodology. A project emphasis was on developing methods that can scale gracefully to large systems. The goals were pursued both on a theory level, developing approaches that reduce hybrid verification problems with temporal logic specifications to more tractable computations that still provide safety and performance guarantees, as well as by developing new computational and simulation algorithms that are principle-based and provide improved efficiency over those currently available. These methods include counter-example guided schemes. Also investigated were fundamental questions about the existence of verification algorithms for certain hybrid system classes. As well, part of the research involved development of new tools for the analysis of hybrid systems that specifically contain the sampling of signals, and the research resulted in significant improvements over existing methods.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 09, 2019
- Accession Number
- AD1085999
Entities
People
- Geir E. Dullerud
- Mahesh Viswanathan
- Matthew P. West
Organizations
- University of Illinois Urbana–Champaign