A Fault Tolerant Scheduling Heuristics for Distributed Real Time Embedded System

Abstract

The integration of a fault tolerance mechanism in critical real-time embedded systems is an important and required property to ensure the continuity of delivering the expected service even in the presence of faults to avoid catastrophic consequences that can be generated in the event of failure of these systems. In this research paper we present a solution to tolerate permanent faults of one processor in heterogeneous distributed real-time embedded systems by using software redundancy solutions based on active and passive replication of dependent tasks in the point-to-point connection. The methodology proposed consists to generate a distribution/scheduling of tasks on hardware architecture and also to tolerate permanent faults of a single processor by executing simultaneously two replicas of a task, the first which ends its execution blocks the second is running. this principle saves very considerable time in distribution/scheduling length with and without errors.