PRESTO: Predicting System-level Disruptions through Parametric Model Checking

TL;DR

PRESTO is a proactive approach that uses parametric model checking and time-series analysis to predict system-level disruptions in self-adaptive systems, enabling earlier mitigation of potential violations.

Contribution

It introduces a novel two-stage method combining time-series analysis and parametric model checking for proactive disruption prediction in self-adaptive systems.

Findings

Effective trend detection in monitoring data

Accurate prediction of future requirement violations

Application demonstrated in autonomous farming domain

Abstract

Self-adaptive systems are expected to mitigate disruptions by continually adjusting their configuration and behaviour. This mitigation is often reactive. Typically, environmental or internal changes trigger a system response only after a violation of the system requirements. Despite a broad agreement that prevention is better than cure in self-adaptation, proactive adaptation methods are underrepresented within the repertoire of solutions available to the developers of self-adaptive systems. To address this gap, we present a work-in-progress approach for the pre diction of system-level disruptions (PRESTO) through parametric model checking. Intended for use in the analysis step of the MAPE-K (Monitor-Analyse-Plan-Execute over a shared Knowledge) feedback control loop of self-adaptive systems, PRESTO comprises two stages. First, time-series analysis is applied to monitoring data in order to identify trends in the values of individual system and/or environment parameters. Next, future non-functional requirement violations are predicted by using parametric model checking, in order to establish the potential impact of these trends on the reliability and performance of the system. We illustrate the application of PRESTO in a case study from the autonomous farming domain.