Adaptive Task Automata with Earliest-Deadline-First Scheduling

Abstract

Adjusting to resource changes, dynamic environmental conditions, or new usage modes are some of the reasons why real-time embedded systems need to be adaptive. This requires a rigorous framework for designing such systems, to ensure that the adaptivity does not result in invalidating the system’s real-time constraints.

To address this need, we have recently introduced adaptive task automata, a framework for modeling, verification, and schedulability analysis in adaptive, hard real-time embedded systems, assuming a fixed-priority scheduler.

In this work, we extend the adaptive task automata framework to incorporate the earliest-deadline-first scheduling policy, as well as enable implementation of any other dynamic scheduling policy. To prove the decidability of our model, and at the same time maintain a manageable degree of conciseness, we show an encoding of our model as a network of timed automata with clock updates. To support this, we also show that reachability in our class of timed automata with updates is decidable. Our contribution helps to streamline the process of designing safety critical adaptive embedded systems.