文献类型：期刊文献

英文题名：Data Collaboration and Task Offloading Strategies for Edge-Vehicle Networks Based on DQN

作者：Xiao, Kaile[1]; Sun, Zhiyuan[1]; Zhao, Haiyan[1]; He, Fangyuan[1]

第一作者：Xiao, Kaile

机构：[1] Beijing Union University, School of Applied Science and Technology, Beijing, China

第一机构：北京联合大学应用科技学院

年份：2025

起止页码：87-91

外文期刊名：2025 7th International Conference on Next Generation Data-Driven Networks, NGDN 2025

收录：EI(收录号：20254319382414)

语种：英文

外文关键词：Data reliability - Energy utilization - Green computing - Intelligent vehicle highway systems - Mobile edge computing - Resource allocation - Vehicles

摘要：With the development of intelligent transportation and Internet of Things technologies, the Edge-Vehicle Network plays a key role in enhancing vehicle computing capabilities and quality of service. This paper focuses on the edge-vehicle network, introduces Deep Q-Network (DQN) methods to achieve efficient data collaboration and task offloading strategies, and proposes the MOTO algorithm. First, this paper constructs a high-reliability model for data collaboration in edge-vehicle networks and derives the delay formula for tasks from mobile vehicle terminals to edge server nodes. Then, tasks are divided into fine-grained categories according to their types and structures. Aiming to minimize delay and under energy consumption constraints, DQN are used to optimize data collaboration and task offloading for vehicles. Finally, by designing the state space, action space, and reward function, the agent can learn optimal resource allocation strategies through continuous exploration and trial-and-error. Experimental results show that the MOTO algorithm has significant advantages over traditional algorithms in reducing task processing delay and improving resource utilization efficiency. ? 2025 IEEE.