EPSRC-funded research seeks to advance the reliability of essential computer systems
Posted: 10th December 2024
Professor of Computer Science, Stefan Kiefer, has been awarded £620k from the Engineering and Physical Sciences Research Council (EPSRC) to advance his research into the foundations of computer-aided verification of probabilistic systems.
Probabilistic systems are systems that include an element of randomness. Computers using these systems are widely employed, such as in aeroplanes or medical devices, and as such their failure can have catastrophic consequences. The accuracy and performance of the systems can be analysed with the help of specialised programs.
Professor Kiefer’s work will focus on the foundations of computer-aided verification of probabilistic systems through tools such as Prism and Storm. He will collaborate closely with Professor Franck van Breugel from York University (Toronto, Canada). They aim to improve the robustness of current verification tools to increase the reliability of verification results, and to develop more broadly applicable algorithms for comparing the behaviour of computer systems.
'I am delighted to receive this EPSRC award. We hope that this research will make the verification of large probabilistic systems more robust. Our approach is based on techniques from various mathematical fields and theoretical computer science, putting them into practice.' Stefan Kiefer, Professor of Computer Science
Even a small system can have a high number of different states, and formal verification tools must minimise the state space (the collection of all possible states a system can be in). Kiefer’s work intends to achieve this by identifying states that behave essentially in the same way, which can then be merged to speed up verification. It also aims to devise new algorithms to improve the diagnosis of systems.
Through his research, Kiefer is set to extend the foundations of probabilistic verification and develop more powerful and more widely applicable verification tools, to improve the correctness and robustness of computer systems that we rely upon every day.