Online Parametric Timed Pattern Matching with Automata-Based Skipping

Timed pattern matching has strong connections with monitoring real-time systems. Given a log and a specification containing timing parameters (that can capture uncertain or unknown constants), parametric timed pattern matching aims at exhibiting for which start and end dates, as well as which parameter valuations, a specification holds on that log. This problem is notably close to robustness. We propose here a new framework for parametric timed pattern matching. Not only we dramatically improve the efficiency when compared to a previous method based on parametric timed model checking, but we further propose optimizations based on skipping. Our algorithm is suitable for online monitoring, and experiments show that it is fast enough to be applied at runtime.