Abstract
Many sensor-based control systems are dynamically changing, and thus require a flexible scheduler. The rate monotonic (RM) real-time scheduling algorithm does not support such dynamic systems very well. On the other hand, with earliest-deadline-first (EDF) and minimum-laxity-first (MLF) dynamic scheduling algorithms, a transient overload in the system may cause a critical task to fail, which is certainly undesirable. This paper proposes a new real-time scheduling algorithm, which we call maximum-urgency-first (MUF), which combines the advantages of the RM, EDF, and MLF algorithms. Like EDF and MLF, MUF has a schedulable bound of 100% for the critical set. And like RM, a critical set can be defined that is guaranteed to meet all its deadlines. The MUF algorithm also allows the scheduler to detect three forms of deadline failures, and call failure handler routines for tasks which fail to meet their deadlines. The MUF scheduler has been implemented as the default scheduler of CHIMERA II, a real-time operating system being used to control sensor-based control systems both at Carnegie Mellon University and elsewhere. There are still many issues to be addressed with regards to the MUF algorithm. This paper also presents those issues, with possible approaches that should be investigated further.