Energy efficient task scheduling for heterogeneous multicore processors in edge computing.

Area of Science:

• Computer Science
• Electrical Engineering
• Energy Systems

Background:

• Edge computing demands efficient resource management on heterogeneous multicore processors (HMPs).
• Existing task scheduling and dynamic voltage and frequency scaling (DVFS) methods lack integrated approaches for energy efficiency.
• Reactive workload adaptation and energy prediction alone do not sufficiently address energy-performance trade-offs in HMPs.

Purpose of the Study:

• To develop a novel algorithm for energy-efficient task scheduling on HMPs in edge computing environments.
• To effectively integrate task prioritization, core-aware mapping, and predictive DVFS.
• To reduce energy consumption while minimizing deadline misses.

Main Methods:

• Proposed a new algorithm combining task prioritization, core-aware mapping, and predictive dynamic voltage and frequency scaling (DVFS).
• Evaluated the algorithm's performance against state-of-the-art methods on real HMP platforms.
• Assessed energy consumption and deadline miss rates under varying workloads.

Main Results:

• Achieved a 20.9% reduction in energy consumption compared to existing methods.
• Maintained a low deadline miss rate of 2.4%.
• Demonstrated scalability and adaptability to diverse workloads on HMP platforms.

Conclusions:

• The proposed algorithm offers a significant advancement in energy-efficient task scheduling for edge computing on HMPs.
• Successfully balances performance requirements (low deadline miss rate) with substantial energy savings.
• Provides a practical and adaptable solution for real-world heterogeneous multicore processor systems.