Scheduling of real time embedded systems for resource and energy minimization by voltage scaling

The aspects of real-time embedded computing are explored with the focus on novel real-time scheduling policies, which would be appropriate for low-power devices. To consider real-time deadlines with pre-emptive scheduling policies will require the investigation of intelligent scheduling heuristics. These aspects for various other RTES models like Multiple processor system, Dynamic Voltage Scaling and Dynamic scheduling are the focus of this thesis. Deadline based scheduling of task graphs representative of real time systems is performed on a multiprocessor system; A set of aperiodic, dependent tasks in the form of a task graph are taken as the input and all the required task parameters are calculated. All the tasks are then partitioned into two or more clusters allowing them to be run at different voltages. Each cluster, thus voltage scaled results in the overall minimization of the power utilized by the system. With the mapping of each task to a particular voltage done, the tasks are scheduled on a multiprocessor system consisting of processors that can run at different voltages and frequencies, in such a way that all the timing constraints are satisfied