TEFRED: A Temperature and Energy Cognizant Fault-Tolerant Real-Time Scheduler Based on Deadline Partitioning for Heterogeneous Platforms

Abstract

AbstractEnergy consumption and peak temperatures on MPSoCs are growing exponentially as transistor density increases, rendering systems unstable. Thus, in modern real-time systems, fault-tolerance is an essential design feature. This paper proposes TEFRED, a heuristic scheduling strategy that addresses the problem of controlling energy and peak temperature levels simultaneously on systems with two types of cores while remaining resistant to transient faults. Our experimental results demonstrate that TEFRED can save considerable energy and lower core peak temperatures compared to a state-of-the-art energy-efficient fault-tolerant scheduler. KeywordsHeterogeneousFault-tolerantTemperatureEnergy