Statistical analysis of execution time profile for temporal validation of a distributed hard real-time system

• Published in: 2021 22nd IEEE International Conference on Industrial Technology (ICIT) Vol. 1; pp. 1188 - 1192
• Main Authors: Prabhakara, Arpitha, Steinwender, Benjamin, Elmenreich, Wilfried
• Format: Conference Proceeding
• Language: English
• Published: IEEE 10.03.2021
• Subjects: Analytical models; Computational modeling; Data models; Fault-Tolerant System; Measurement-based Probabilistic Timing Anal-ysis; Probabilistic logic; Probability Distribution; Real-Time Distributed System; Real-time systems; Semiconductor device measurement; Statistical analysis; Timing Failure; WCET Analysis
• DOI: 10.1109/ICIT46573.2021.9453493

This paper addresses modelling and assessing of worst-case execution times of tasks in a distributed real-time system. The presented case study is based on a stress test system for power semiconductors.Typical distributed systems involve inherent inter-dependencies that have to be handled with minimized delays. Studying the computing task's timing behavior becomes necessary to understand the timing violations. Modelling the temporal behaviour of the system using probabilistic Worst-Case Execution Time (WCET) analysis aids to overcome the timing faults.This paper focuses on (1) the study of the temporal behaviour through the execution time profile of the distributed hard real-time system; (2) the statistical analysis by performing probability distribution modelling on the measured data. Measurement-based probabilistic timing analysis is an emerging and reliable method used to arrive at industry quality estimates. This method is used here in the paper to carry out temporal validation of the real-time computing tasks on our case study.