Value Based Overload Handling of Aperiodic Tasks in Offline Scheduled Real-Time Systems
Jan Carlson, Tomas Lennvall, Gerhard Fohler · 2001
This paper describes a runtime scheduling method for a mix of offline scheduled and value based dynamic tasks with a focus on overload situations. We present a formulation of overload handling as a general binary optimisation problem and give an algorithm for solving it. Our method is based on slot shifting [4] to provide for the integration of offline and online scheduling. Dynamic tasks are scheduled according to basic EDF, extended with an algorithm for overload detection and resolution. We include a penalty value for tasks which have been guaranteed but missed their deadline, e.g., due to rejection under overload