BROWSE
Detail
Journal Article
Runtime verification of cause–effect latency in black-box systems
- Authors
- Yong-Jun Shin
- Issue Date
- 2026-04
- Citation
- Information and Software Technology, v.196, pp.1-16
- ISSN
- 0950-5849
- Publisher
- Elsevier
- Language
- English
- Type
- Journal Article
- Abstract
- Context: Safety-critical systems such as autonomous driving rely on cause–effect chains of asynchronous tasks, where end-to-end latency must meet strict timing constraints. While design-time analysis provides guarantees, runtime verification is required to handle runtime uncertainties. However, existing runtime methods often rely on intrusive instrumentation or detailed white-box access, which are rarely available in practice. Objective: This study aims to enable runtime verification of cause–effect latency in black-box systems, where only task output events (write-events) are observable. Methods: We formally extend the classic cause–effect latency verification problem to the black-box setting and propose two lightweight algorithms: (1) a job-chain estimation algorithm that safely over-estimates latency based on limited observations, and (2) a statistical verification algorithm that produces verdicts under user-specified confidence levels, avoiding false positives by design. Results: We theoretically prove the over-estimation error bounds and empirically validate them through controlled experiments. Results show that estimation errors inherent to limited observability diminish when chain utilization is high, the verification algorithm remains trustworthy with extremely low false positives, while exhibiting limited conservative false negatives as a trade-off, and both algorithms incur negligible runtime cost. An industrial case study on an autonomous driving system further confirms the practicality of the proposed approach, successfully verifying 39 chains under black-box constraints with minimal engineering effort. Conclusion: The findings demonstrate that black-box runtime verification of cause–effect latency is both feasible and effective, providing a lightweight and practical foundation for verifying timing requirements in complex, safety-critical systems.
- Keyword
- Runtime verification, Cause–effect latency, End-to-end latency, Cause–effect chain, Black-box systems
- KSP Keywords
- Autonomous driving system, Black box, Box constraints, Confidence levels, Controlled experiments, Design time, End to End(E2E), Estimation error, False Positive(FP), False negative, Industrial Case Study
This work is distributed under the term of Creative Commons License (CCL)
(CC BY NC ND)
(CC BY NC ND)
