Reliability guarantees in automata-based scheduling for embedded control software

Automata-based scheduling is a recent technique for online scheduling of software control components in embedded systems. This letter studies one important aspect of automata-based scheduling that has not been studied in the past, namely resilience to faults. The goal of the proposed technique is to create an automaton that recommends the scheduling patterns that are admissible with respect to control performance requirements, when the state of the system has been mutated by faults. The problem has been formulated as a game between the scheduler and the (possibly faulty) system, where a winning strategy of the scheduler prevents the system from reaching bad states forever. We present a method for analyzing the structure of the game and extracting an automaton that captures the winning strategies of the scheduler, namely the automaton to be used for automata-based scheduling