A value distributional deep reinforcement learning framework for intelligent offloading in end–edge–cloud computing

doi:10.1016/j.comnet.2026.112569

A value distributional deep reinforcement learning framework for intelligent offloading in end–edge–cloud computing

2026-07-16

0

PRE

AI

DOI：10.1016/j.comnet.2026.112569

原文链接

原文求助

摘要

En 中文

• Propose QD3QN-PER combining quantile regression, double dueling DQN, and PER for offloading. • Model full return distribution to improve decision accuracy under uncertainty. • Employ prioritized experience replay to boost sample efficiency and learning diversity. • Use double dueling DQN to reduce value estimation bias and enhance stability. • Demonstrate superior task offloading performance in end–edge–cloud collaborative network.

Keyword:

Deep reinforcement learning

Mobile edge computing

Computation offloading

Prioritized experience replay

End–edge–cloud collaboration

A value distributional deep reinforcement learning framework for intelligent offloading in end–edge–cloud computing

摘要

期刊

机构

引用论文