Clairvoyant Monitoring for Signal Temporal Logic

• Published in: Formal Modeling and Analysis of Timed Systems Vol. 12288; pp. 178 - 195
• Main Authors: Qin, Xin, Deshmukh, Jyotirmoy V.
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 01.01.2020; Springer International Publishing
• Series: Lecture Notes in Computer Science
• ISBN: 9783030576271; 3030576272
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-57628-8_11

In this paper, we consider the problem of monitoring temporal patterns expressed in Signal Temporal Logic (STL) over time-series data in a clairvoyant fashion. Existing offline or online monitoring algorithms can only compute the satisfaction of a given STL formula on the time-series data that is available. We use off-the-shelf statistical time-series analysis techniques to fit available data to a model and use this model to forecast future signal values. We derive the joint probability distribution of predicted signal values and use this to compute the satisfaction probability of a given signal pattern over the prediction horizon. There are numerous potential applications of such prescient detection of temporal patterns. We demonstrate practicality of our approach on case studies in automated insulin delivery, unmanned aerial vehicles, and household power consumption data.