QTWTL: Quality Aware Time Window Temporal Logic for Performance Monitoring

TL;DR

This paper introduces QTWTL, a new logic for performance monitoring in service-oriented applications, extending TWTL with counting and aggregation, and provides efficient offline algorithms tested on real-world taxi data.

Contribution

The paper presents QTWTL, a novel extension of TWTL with counting and aggregation, along with scalable offline monitoring algorithms for performance analysis.

Findings

Monitoring algorithm has linear space complexity.

Monitoring algorithm has linear time complexity.

Effective performance specification and monitoring on real-world data.

Abstract

In various service-oriented applications such as distributed autonomous delivery, healthcare, tourism, transportation, and many others, where service agents need to perform serial and time-bounded tasks to achieve their goals, quality of service must constantly be assured. In addition to safety requirements, such agents also need to fulfill performance requirements in order to satisfy their quality of service. This paper proposes the novel quality-aware time window temporal logic (QTWTL) by extending the traditional time window temporal logic (TWTL) with two operators for counting and aggregation operations. We also propose offline runtime monitoring algorithms for the performance monitoring of QTWTL specifications. To analyze the feasibility and efficiency of our proposed approach, we generate a large number of traces using the New York City Taxi and Limousine Commission Trip Record data, formalize their performance requirements using QTWTL, and monitor them using the proposed algorithms. The obtained results show that the monitoring algorithm has a linear space and time complexity with respect to the number of traces monitored.