基于图神经网络和深度强化学习的车联网边缘计算卸载算法

doi:10.6040/j.issn.1672-3961.0.2025.205

摘要/Abstract

摘要： 为了在应用边缘计算时充分利用车联网系统内的资源,降低处理任务所需时延、能耗及处理成本,提出一种基于图神经网络和深度强化学习的车联网边缘计算卸载算法。将计算任务卸载问题建模为马尔可夫决策过程,最小化任务的平均时延与能耗;利用图神经网络的全局特征学习能力,将原本独立的车辆个体联系为一个整体,使每辆车都能做出全局最优的卸载决策。仿真结果表明,与本地计算、完全卸载、随机决策、双深度Q网络算法相比,所提算法的平均时延分别降低约88.9%、49.9%、27.6%和9.4%,平均成本分别降低约74.5%、46.1%、26.3%和6.2%;与完全卸载、随机决策、双深度Q网络算法相比,所提算法的平均能耗分别降低约43.6%、25.4%和4.2%。

关键词: 车联网, 边缘计算, 任务卸载, 图神经网络, 深度强化学习

Abstract: To fully utilize the resources within the Internet of Vehicles system when applying edge computing, and reduce the delay, energy consumption and processing costs required for task processing, an offloading algorithm in edge computing for Internet of Vehicles based on graph neural network and deep reinforcement learning was proposed. The task offloading problem was modeled as a Markov decision process to minimize the average delay and energy consumption for task processing. By leveraging the global feature learning capability of graph neural network, the originally independent vehicle entities were integrated into a cohesive whole, enabling each vehicle to make rational offloading decisions. Simulation results showed that, compared with local computing, full offloading, random decision, and double deep Q-network, the proposed algorithm reduced average delay by approximately 88.9%, 49.9%, 27.6%, and 9.4%, respectively, and reduced average costs by approximately 74.5%, 46.1%, 26.3%, and 6.2%, respectively. Compared with full offloading, random decision, and double deep Q-network, the proposed algorithm reduced average energy consumption by approximately 43.6%, 25.4%, and 4.2%, respectively.

Key words: Internet of Vehicles, edge computing, task offloading, graph neural network, deep reinforcement learning

中图分类号:

TP393

王倩,李明津,孟宪静,耿蕾蕾. 基于图神经网络和深度强化学习的车联网边缘计算卸载算法[J]. 山东大学学报 (工学版), 2026, 56(4): 17-26.

Wang Qian, Li Mingjin, Meng Xianjing, Geng Leilei. Offloading algorithm in edge computing for Internet of Vehicles based on graph neural network and deep reinforcement learning[J]. Journal of Shandong University(Engineering Science), 2026, 56(4): 17-26.

参考文献

[1] 许小龙, 方子介, 齐连永, 等. 车联网边缘计算环境下基于深度强化学习的分布式服务卸载方法[J]. 计算机学报, 2021, 44(12): 2382-2405. Xu Xiaolong, Fang Zijie, Qi Lianyong, et al. A deep reinforcement learning-based distributed service offloading method for edge computing empowered Internet of vehicles[J]. Chinese Journal of Computers, 2021, 44(12): 2382-2405.
[2] 许世琳. 车联网中基于深度强化学习的计算任务卸载策略研究[D]. 北京: 北京邮电大学, 2021: 1-2. Xu Shilin. Research on computation offloading strategy based on deep reinforcement learning in vehicular networks[D]. Beijing: Beijing University of Posts and Telecommunications, 2021: 1-2.
[3] Yang J M, Shah A A, Pezaros D. A survey of energy optimization approaches for computational task offloading and resource allocation in MEC networks[J]. Elec-tronics, 2023, 12(17): 3548.
[4] Zhao L, Li T Y, Zhang E C, et al. Adaptive swarm intelligent offloading based on digital twin-assisted prediction in VEC[J]. IEEE Transactions on Mobile Computing, 2024, 23(8): 8158-8174.
[5] Chen Y, Zhao F J, Chen X, et al. Efficient multi-vehicle task offloading for mobile edge computing in 6G networks[J]. IEEE Transactions on Vehicular Technology, 2022, 71(5): 4584-4595.
[6] Liu J, Wang S B, Wang J T, et al. A task oriented computation offloading algorithm for intelligent vehicle network with mobile edge computing[J]. IEEE Access, 2019, 7: 180491-180502.
[7] Huang X M, Yu R, Ye D D, et al. Efficient workload allocation and user-centric utility maximization for task scheduling in collaborative vehicular edge computing[J]. IEEE Transactions on Vehicular Technology, 2021, 70(4): 3773-3787.
[8] Ku Y J, Baidya S, Dey S. Adaptive computation partitioning and offloading in real-time sustainable vehicular edge computing[J]. IEEE Transactions on Vehicular Technology, 2021, 70(12): 13221-13237.
[9] Bute M S, Fan P Z, Zhang L, et al. An efficient distributed task offloading scheme for vehicular edge computing networks[J]. IEEE Transactions on Vehicular Technology, 2021, 70(12): 13149-13161.
[10] Deng Y Q, Chen Z G, Yao X, et al. Parallel offloading in green and sustainable mobile edge computing for delay-constrained IoT system[J]. IEEE Transactions on Vehicular Technology, 2019, 68(12): 12202-12214.
[11] Lang P, Tian D X, Duan X T, et al. Cooperative computation offloading in blockchain-based vehicular edge computing networks[J]. IEEE Transactions on Intelligent Vehicles, 2022, 7(3): 783-798.
[12] Hossain M D, Sultana T, Hossain M A, et al. Dynamic task offloading for cloud-assisted vehicular edge computing networks: a non-cooperative game theoretic approach[J]. Sensors, 2022, 22(10): 3678.
[13] Alam M Z, Jamalipour A. Multi-agent DRL-based Hungarian algorithm(MADRLHA)for task offloading in multi-access edge computing Internet of Vehicles(IoVs)[J]. IEEE Transactions on Wireless Communi-cations, 2022, 21(9): 7641-7652.
[14] Lu J R, Chen L Y, Xia J J, et al. Analytical offloading design for mobile edge computing-based smart Internet of Vehicle[J]. EURASIP Journal on Advances in Signal Processing, 2022, 2022(1): 44.
[15] Zhu X Y, Luo Y Y, Liu A F, et al. Multiagent deep reinforcement learning for vehicular computation offloading in IoT[J]. IEEE Internet of Things Journal, 2021, 8(12): 9763-9773.
[16] Zhang D G, Cao L X, Zhu H L, et al. Task offloading method of edge computing in Internet of Vehicles based on deep reinforcement learning[J]. Cluster Computing, 2022, 25(2): 1175-1187.
[17] Pan C, Wang Z, Zhou Z Y, et al. Deep reinforcement learning-based URLLC-aware task offloading in collaborative vehicular networks[J]. China Communi-cations, 2021, 18(7): 134-146.
[18] Shi J M, Du J, Wang J J, et al. Priority-aware task offloading in vehicular fog computing based on deep reinforcement learning[J]. IEEE Transactions on Vehicular Technology, 2020, 69(12): 16067-16081.
[19] Wu Z Y, Yan D F. Deep reinforcement learning-based computation offloading for 5G vehicle-aware multi-access edge computing network[J]. China Communications, 2021, 18(11): 26-41.
[20] Wang K, Wang X F, Liu X. Sustainable Internet of Vehicles system: a task offloading strategy based on improved genetic algorithm[J]. Sustainability, 2023, 15(9): 7506.
[21] 曹宇慧, 黄昱泽, 冯北鹏, 等. 基于深度强化学习的物联网服务协同卸载方法[J]. 山东大学学报(工学版), 2024, 54(1): 83-90. Cao Yuhui, Huang Yuze, Feng Beipeng, et al. A collaborative service offloading approach for Internet of Things based on deep reinforcement learning[J]. Journal of Shandong University(Engineering Science), 2024, 54(1): 83-90.
[22] Kong X J, Yang X X, Shen S, et al. Energy-delay joint optimization for task offloading in digital twin-assisted Internet of Vehicles[J]. ACM Transactions on Sensor Networks, 2024: 3658671.
[23] Shu W N, Yu H X, Zhai C, et al. An adaptive computing offloading and resource allocation strategy for Internet of Vehicles based on cloud-edge collaboration[J]. IEEE Transactions on Intelligent Transportation Systems, 2025, 26(10): 17349-17358.
[24] Parekh A K, Gallager R G. A generalized processor sharing approach to flow control in integrated services networks: the single node case[J]. IEEE/ACM Transactions on Networking, 1994, 2(2): 137-150.
[25] 王伟平. 边缘计算网络中的任务卸载与资源分配技术研究[D]. 北京: 北方工业大学, 2024: 25. Wang Weiping. Research on task offloading and resource allocation technologies in edge computing networks[D]. Beijing: North China University of Technology, 2024: 25.
[26] Wang Z Y, Schaul T, Hessel M, et al. Dueling network architectures for deep reinforcement learning[C] //Proceedings of the 33rd International Conference on Machine Learning. New York, USA: JMLR, 2016: 1995-2003.
[27] Van Hasselt H, Guez A, Silver D. Deep reinforcement learning with double Q-learning[C] //Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence. Phoenix, USA: AAAI, 2016: 2094-2100.

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed