A Compositional Approach to Diagnosing Faults in Cyber-Physical Systems

TL;DR

This paper presents a compositional method for diagnosing faults in cyber-physical systems by leveraging assume-guarantee contracts, enabling efficient fault localization at the component level.

Contribution

It introduces a systematic approach using contract-based design to identify faulty components and specific predicates responsible for failures in CPS.

Findings

Successfully applied to autonomous vehicle scenarios.

Efficient identification of faulty components and predicates.

Demonstrated effectiveness with illustrative examples.

Abstract

Identifying the cause of a system-level failure in a cyber-physical system (CPS) can be like tracing a needle in a haystack. This paper approaches the problem by assuming that the CPS has been designed compositionally and that each component in the system is associated with an assume-guarantee contract. We exploit recent advances in contract-based design that show how to compute the contract for the entire system using the component-level contracts. When presented with a system-level failure, our approach is able to efficiently identify the components that are responsible for the system-level failure together with the specific predicates in those components' specifications that are involved in the fault. We implemented this approach using Pacti and demonstrate it through illustrative examples inspired by an autonomous vehicle in the DARPA urban challenge.