Adaptive fault-tolerance with statically scheduled real-time systems

Safety critical real-time systems need to complement their methods for meeting timing constraints with fault tolerance for real-world applications. While scheduling algorithms to provide fixed reliability levels have been introduced, methods to enable a real-time system to adapt to changing reliability requirements at run-time have been given little attention. In this paper we present an approach for adaptive fault tolerance with statically scheduled real-time systems. The idea is to provide a minimum reliability level by constructing a static schedule. At run-time, we use a method to integrate static and dynamic scheduling, slot shifting. It is used to incorporate a variety of fault tolerance activities into the static schedule dynamically without impairing the feasibility of the pre-guaranteed tasks and minimum reliability activities. Our algorithms work in distributed systems with precedence constrained tasks, allowing the application of time and space redundant fault tolerance techniques with high resource utilization.