Temporal and Spatial Fault Detection for Connected Cyber-Physical Systems

TY - CHAP

T1 - Temporal and Spatial Fault Detection for Connected Cyber-Physical Systems

AU - Araujo, Hugo

AU - Mousavi, Mohammadreza

AU - Nejati, Shiva

PY - 2025/4/28

Y1 - 2025/4/28

N2 - Testing connected cyber-physical systems (CPS) is a com-plex task. Connected CPS feature complex stochastic dynamic behaviourin interaction with the physical and human environment as well as com-munication over networks. Devising an oracle for testing connected CPSis a challenge; the oracle should be able to quantitatively reason aboutthe stochastic nature of the interactions between the CPS and its en-vironment. The quantitative reasoning should be sensitive to significantdeviations in the dynamics and neglect minor deviations, e.g., due tomeasurement errors. To address this challenge, we provide the mathe-matical framework for conformance testing of connected CPS. We definea quantitative measure of closeness for two distributions of trajectories(i.e., output distributions from two distinct stochastic systems that areprovided with the same input stimuli) that allows for capturing signif-icant temporal and spatial deviations and neglecting subtle ones. Thismeasure forms the basis for our notion of stochastic conformance, whichdetermines when two stochastic systems conform to each other. We im-plement our proposed notion of stochastic conformance and compare ournotion against a state-of-the-art baseline by applying both approachesto a case study involving a platoon of connected vehicles. Our notion de-tects a variety of different types of faults whilst allowing subtle deviationsresulting from naturally occurring perturbations inherent to CPS.

AB - Testing connected cyber-physical systems (CPS) is a com-plex task. Connected CPS feature complex stochastic dynamic behaviourin interaction with the physical and human environment as well as com-munication over networks. Devising an oracle for testing connected CPSis a challenge; the oracle should be able to quantitatively reason aboutthe stochastic nature of the interactions between the CPS and its en-vironment. The quantitative reasoning should be sensitive to significantdeviations in the dynamics and neglect minor deviations, e.g., due tomeasurement errors. To address this challenge, we provide the mathe-matical framework for conformance testing of connected CPS. We definea quantitative measure of closeness for two distributions of trajectories(i.e., output distributions from two distinct stochastic systems that areprovided with the same input stimuli) that allows for capturing signif-icant temporal and spatial deviations and neglecting subtle ones. Thismeasure forms the basis for our notion of stochastic conformance, whichdetermines when two stochastic systems conform to each other. We im-plement our proposed notion of stochastic conformance and compare ournotion against a state-of-the-art baseline by applying both approachesto a case study involving a platoon of connected vehicles. Our notion de-tects a variety of different types of faults whilst allowing subtle deviationsresulting from naturally occurring perturbations inherent to CPS.

M3 - Conference paper

T3 - Lecture Notes in Computer Science

BT - Proceedings of the 45th International Conference on Formal Techniques for Distributed Objects, Components, and Systems

ER -