An off-line multiprocessor real-time scheduling algorithm to reduce static energy consumption

Vincent Legout (1, 2, 3) , Mathieu Jan (1) , Laurent Pautet (2, 3)

1 LIST (CEA) - Laboratoire d'Intégration des Systèmes et des Technologies
2 LTCI - ACES - Autonomic and Critical Embedded Systems
3 INFRES - Département Informatique et Réseaux

Energy consumption of highly reliable real-time embedded systems is a significant concern. Static energy consumption tends to become more important than dynamic energy consumption. This paper aims to propose a new off-line scheduling algorithm to put as much as possible processors in low- power states instead of idling. In these states, energy consumption is reduced, enhancing the battery life-time of mission critical systems. However, no instruction can be executed and a transition delay is required to come back to the active state. Activating deeper low-power states requires to produce larger idle periods. As the processor usage is constant for a given task set, this objective implies reducing the number of idle periods. Our proposal is to modelize the processors idle time as an additional task. Then we formalize the problem as a linear equation system with the objective of reducing the number of preemptions (or executions) of this additional task. Simulations show that our algorithm is more energy efficient than existing algorithms.

Domains: Computer Science [cs]; Distributed, Parallel, and Cluster Computing [cs.DC]; Computer Science [cs]; Embedded Systems; Computer Science [cs]; Data Structures and Algorithms [cs.DS]

Vincent Legout, Mathieu Jan, Laurent Pautet. An off-line multiprocessor real-time scheduling algorithm to reduce static energy consumption. First Workshop on Highly-Reliable Power-Efficient Embedded Designs (HARSH), Feb 2013, Shenzhen, China. pp.7-12. ⟨hal-00919169⟩