Energy-aware stochastic scheduling model with precedence constraints on DVFS-enabled processors

The stochastic scheduling of precedence-constrained jobs on a heterogeneous processor is a challenging problem that requires solutions with one or more optimized QoS parameters. In this work, an energy-aware stochastic algorithm is proposed to schedule the batch of precedence-constrained jobs on heterogeneous DVFS-enabled processors with the objective of optimizing turnaround time and energy consumption. The processing time of tasks in all jobs and their precedence-constraint times are governed by independent probability distributions. The performance of the proposed stochastic algorithm is compared with SHEFT and ECS based on randomly generated batches of different sizes. The experimental study reveals that the proposed algorithm significantly outperforms the SHEFT and ECS algorithms in terms of turnaround time and energy consumption.