Utilization and criticality based fault-tolerant scheduling in multicore mixed critical systems

What is it about?

Mixed critical systems are real-time systems where each task has a different criticality level. All modern complex embedded applications like industrial control systems, flight management systems, autopilot systems, etc. fall in this category. The tasks in such systems should not miss their deadlines. The literature reveals that research has concentrated on achieving fault tolerance in multiprocessor real-time systems. The investigation of scheduling in complex mixed systems to achieve multiple objectives is missing. The main purpose of this work is to investigate the task scheduling mechanisms in mixed critical systems on multicore architecture where multiple objectives are the performance measures. The real-time patch of Linux is used for simulation of the mixed critical task sets for experimentation purposes on a quad-core homogenous architecture. The performance of priority-based scheduling mechanisms is discussed based on the simulation of an air traffic control system case study and 50 random task sets. To achieve the performance measures a partitioning mechanism is proposed based on the first-fit algorithm. This proposed technique with the earliest deadline first virtual deadlines achieves an improvement in the success ratio of the tasks by 11% as compared to the global approach. The proposed partitioning algorithm utilizes only three processor cores and reliability is achieved by replication of high critical tasks. The investigation results can be used to achieve fault tolerance, improve schedulability and reduce the cores utilized in any of the mixed critical systems.

Featured Image