Energy aware task scheduling with task synchronization for embedded real time systems

Abstract

Abstract-Slowdown factors determine the extent of slowdown that a computing system can experience based on functional and performance requirements. Dynamic voltage scaling (DVS) of a processor based on slowdown factors can lead to considerable energy savings. This paper addresses the problem of DVS in the presence of task synchronization. Tasks synchronize to enforce mutually exclusive access to the shared resources and can be blocked by lower priority tasks. Task slowdown factors that guarantee meeting all task deadlines are computed. Both static and dynamic priority scheduling viz. rate monotonic (RM) scheduling and earliest deadline first (EDF) scheduling, respectively, are studied.

Energy-efficient scheduling of real-time tasks with shared resources

“…As a result, tasks can be scheduled without violating their timing constraints and the energy consumption could be reduced. Such a strategy is called two-speed strategy (TSS), and it is common for scheduling of real-time tasks with shared resources, e.g., [3,4,5,6,9,10,11,12,13].…”

“…When critical sections are considered to be preemptible, Jejurikar and Gupta [9] proposed a TSS-based scheduling algorithm for fixed-priority tasks, called critical section maximum speed (CSMS), based on the rate monotonic (RM) algorithm [7] and the priority ceiling protocol (PCP) [15]. CSMS always executes tasks at a low speed and then it switches to the maximum processor speed when tasks being executed in a critical section.…”

Model-driven software synthesis for hard real-time applications with energy constraints

“…The work described in [12] proposes a runtime method for dealing with exclusion relations. Nevertheless, precedence relations are not taken into account and preemption as well as voltage/frequency switching overheads are considered in tasks' worst-case execution time (WCET).…”

“…18), τ 2 started at 0, τ 1 at 1.02 and τ 3 at 4.01, τ 4 preempted τ 3 at 5.01 and τ 3 returned at 7.01. Lastly, τ 6 started at 12 and τ 5 at 14. The reader should note that the idle time (from 0 to 2) at beginning of the feasible schedule is located at the end of oscilloscope timing diagram, since the pulse of 2.55 V is generated immediately before the first task instance.…”

Energy minimization for real-time systems with non-convex and discrete operation modes

“…Convexity has a number of profound ramifications when energy is minimized using variable voltage strategies. Several researches have been performed to solve the task-scheduling problem on DVS-enabled systems which focus on dynamic energy reduction and ignore leakage energy [16] [7]. In [5] [17] energy efficient scheduling of periodic real time tasks in a system with DVS processor and multiple non-DVS devices is explored.…”