Using imprecise computation techniques for power management in real-time embedded systems

Embedded systems present severe limitations in terms of processing and memory capabilities and are often powered by batteries, making energy an important resource to be managed. This work explores energy as a parameter for Quality of Service (QoS) of embedded systems. The goal is to guarantee the battery lifetime specified by the application and yet preserve the deadlines of essential (hard realtime) tasks. We propose equations to check at project-time if a given set of tasks are schedulable. At execution-time, a preemptive scheduler for imprecise tasks based on the EDF algorithm prevents the optional subtasks execution when ever there is the possibility of deadline loss or battery exhaustion. A prototype was developed in EPOS using power management mechanisms provided by the system.