Priority-Aware Task Scheduling in Computing Power Network-Enabled Edge Computing Systems

doi:10.1109/TNSE.2025.3557385

Priority-Aware Task Scheduling in Computing Power Network-Enabled Edge Computing Systems

2025-07-01

0

PRE

AI

DOI：10.1109/TNSE.2025.3557385

Original

Original request for help

Save

Abstract

En 中文

The Internet of everything, a potential direction for the next-generation Internet, positions edge collaboration as a promising computing paradigm to address the workload dispersion and resource constraints inherent in traditional edge computing frameworks. However, the increasing complexity of cross-domain networks introduces challenges for efficient task execution and balanced resource utilization in edge collaboration, which remain insufficiently explored. To address these challenges, a next-generation network architecture, the compute power network (CPN), was recently proposed. The CPN leverages ubiquitous connections among heterogeneous resources to optimize task scheduling collaboratively. Building on this concept, we design an edge computing system that integrates CPN to enable dynamic and collaborative task scheduling. Inspired by the sliding window, we develop a dynamic scheduling scheme that prioritizes computing tasks and matches tasks to computing resources in real time. Additionally, we propose an improved deep reinforcement learning (DRL) algorithm to optimize scheduling policies, aiming to improve task success rates, minimize execution delays, and ensure balanced and efficient resource utilization. Lastly, simulation experiments validate the effectiveness of the proposed scheme and algorithm.

Keywords:

Edge cooperation

computing power networks

scheduling priority

load-balancing

delay-sensitive

Priority-Aware Task Scheduling in Computing Power Network-Enabled Edge Computing Systems

Abstract

Journal

Organization