In response to the issue of low computational performance of edge computing systems when dealing with discrete-continuous hybrid decision-making problems in vehicular network edge computing scenarios, a distributed multi-stage decision-making strategy for dynamic offloading of vehicular network edge computing tasks based on deep reinforcement learning is proposed. By dividing the task processing procedure into two sub-problems: offloading decisions and resource allocation, an algorithm framework capable of handling both discrete and continuous action domains was designed based on the D3DQN network and TD3 network. In order to further enhance effective search in the action space and improve algorithm performance, an optimization and dynamic priority update mechanism for search in the action space was designed. Finally, based on the above algorithm framework and optimization mechanisms, a multi-stage decision-making algorithm for dynamic offloading of vehicle network edge computing tasks was developed and implemented. This algorithm can effectively solve discrete-continuous hybrid decision-making problems in vehicle network edge computing scenarios, utilize edge computing system resources, and ensure that vehicles can complete the offloading of computing tasks with minimal delay and energy consumption, according to the current network status and task size. Simulation results show that, compared with baseline offloading methods, the proposed method has faster convergence and lower latency energy consumption. It can make better use of the computing resources of the edge system to provide computing services for vehicle edge task requests, thus further enhancing the quality of service for users.
将固定任务数量情况下不同任务卸载方法和基线卸载方法的时延能耗加权值之差与基线卸载方法的时延能耗加权值的比值,定义为时延能耗缩减率(Energy and delay sum reduction ratio, EDSRR),该指标值越大,说明边缘计算任务卸载系统的性能越好。本文以本地执行方法作为基线卸载方法,通过计算本文方法和其他3种方法与本地执行方法之间的时延能耗缩减率,来进一步验证本文方法的有效性。如图14所示,Double DQN和DDPG时延能耗缩减率都达到了80%以上,本文方法达到了90.1%,比Double DQN提高了近6%,优于其他调度方法。
KuangBo-yu, LiYu-ze, GuFang-ming, et al. Review of internet of vehicle security research: threats, countermeasures, and future prospects[J]. Journal of Computer Research and Development, 2023, 60(10): 2304-2321.
[4]
BobanM, KousaridasA, ManolakisK, et al. Connected roads of the future: use cases, requirements, and design considerations for vehicle-to-everything communications[J]. IEEE Vehicular Technology Magazine, 2018, 13(3): 110-123.
[5]
YangD G, JiangK, ZhaoD, et al. Intelligent and connected vehicles: current status and future perspectives[J]. Science China Technological Sciences, 2018, 61: 1446-1471.
[6]
ZhangW Y, ZhangZ J, chaoH C. Cooperative fog computing for dealing with big data in the internet of vehicles: architecture and hierarchical resource management[J]. IEEE Communications Magazine, 2017, 55(12): 60-67.
[7]
JaehwanL, WoongsooN. A survey on vehicular edge computing architectures[C]∥2022 13th International Conference on Information and Communication Technology Convergence(ICTC), Jeju, Island, Korea, 2022: 2198-2200.
[8]
MaoY Y, YouC S, ZhangJ, et al. A survey on mobile edge computing: the communication perspective[J]. IEEE Communications Surveys & Tutorials, 2017, 19(4): 2322-2358.
[9]
TangM, WongV W S. Deep reinforcement learning for task offloading in mobile edge computing systems[J]. IEEE Transactions on Mobile Computing, 2022, 21(6): 1985-1997.
[10]
ZhaoN, LiangY C, NiyatoD, et al. Deep reinforcement Learning for user association and resource allocation in heterogeneous cellular networks[J]. IEEE Transactions on Wireless Communications, 2019, 18(11): 5141-5152.
[11]
DaiY Y, XuD, ZhangK, et al. Deep reinforcement learning for edge computing and resource allocation in 5G beyond[C]∥2019 IEEE 19th International Conference on Communication Technology, Xi'an, China, 2019: 866-870.
[12]
Raeisi-varzanehM, DakkakO, HabbalA, et al. Resource scheduling in edge computing: architecture, taxonomy, open issues and future research directions[J]. IEEE Access, 2023, 11: 25329-25350.
[13]
FengJ, YuF R, PeiQ Q, et al. Joint optimization of radio and computational resources allocation in blockchain-enabled mobile edge computing systems[J]. IEEE Transactions on Wireless Communications, 2020, 19(6): 4321-4334.
LiZhi-yong, WangQi, ChenYi-fan, et al. A survey on task offloading research in vehicular edge computing[J]. Chinese Journal of Computers, 2021, 44(5): 963-982.
XuBin, ZhaoYun-kai, ZhuJian-ming, et al. Continuous offloading and resource allocation method of uncertain tasks in mobile edge computing[J]. Journal of Software, 2023, 35(3): 1466-1484.
[18]
LiuG Z, DaiF, HuangB, et al. A collaborative computation and dependency-aware task offloading method for vehicular edge computing: a reinforcement learning approach[J]. Journal of Cloud Computing, 2022, 11(1): 1-15.
SunYan-jing, YuZheng-da, ChenRui-rui, et al. Deep reinforcement learning based high reliability resource allocation algorithm for internet of vehicles[J]. Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition), 2023, 35(4): 706-714.
[21]
HaoH, DingW, ZhangW. Time-continuous computing offloading algorithm with user fairness guarantee[J]. Journal of Network and Computer Applications, 2024, 223: 103826.
[22]
YanJ, BiS Z, ZhangY J A. Offloading and resource allocation with general task graph in mobile edge computing: a deep reinforcement learning approach[J]. IEEE Transactions on Wireless Communications, 2020, 19(8): 5404-5419.
[23]
ZhaoJ H, LiQ P, GongY, et al. Computation offloading and resource allocation for cloud assisted mobile edge computing in vehicular networks[J]. IEEE Transactions on Vehicular Technology, 2019, 68(8): 7944-7956.
[24]
HanH, WuD G, ZhouF H, et al. Dynamic task offloading in MEC-enabled IoT networks: a hybrid DDPG-D3QN approach[C]∥2021 IEEE Global Communications Conference(GLOBECOM). Madrid, Spain, 2021: 1-6.
[25]
GuanT, YaoY S, ChaoY, et al. Energy-efficient computing offloading policy for MEC-assisted power sensor network: a D3QN approach[C]∥2023 International Conference on Cyber-Physical Social Intelligence(ICCSI), Xi'an, China, 2023: 180-185.
[26]
WangZ Y, SchaulT, HesselM, et al. Dueling Network Architectures for Deep reinforcement learning[C]∥JMLR.org. Proceedings of the 33rd International Conference on International Conference on Machine Learning-Volume 48, New York, NY, USA, 2016: 1995-2003.
[27]
HasseltH V, GuezA, SilverD. Deep reinforcement learning with double Q-learning[C]∥AAAI'16. Proceeding of the Thirtieth AAAI Conference on Artificial Intelligence, Phoenix, Arizona, 2016: 2094-2100.
ZhangHong-Xia, LvZhi-hao, XiShi-yu, et al. A method for offloading vehicle collaborative tasks for green computing[J]. Journal of Electronics & Information Technology, 2024, 48(1): 175-183.
[32]
KiranB R, SobhI, TalpaertV, et al. Deep reinforcement learning for autonomous driving: A survey[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(6): 4909-4926.
[33]
ZhouD R, LiL H, GuQ Q. Neural contextual bandits with ucb-based exploration[C]∥International Conference on Machine Learning. 2020: 11492-11502.
[34]
ZhongJ W, ChenX, JiaoL B. A fair energy-efficient offloading scheme for heterogeneous fog-enabled IoT using D3QN[C]∥2023 25th International Conference on Communication Software and Networks(ICCSN), IEEE, 2023: 47-52.
[35]
MinM H, XiaoL, ChenY, et.al. Learning-based computation offloading for IoT devices with energy harvesting[J]. IEEE Transactions on Vehicular Technology, 2019, 68(2): 1930-1941.
[36]
GengL W, ZhaoH B, WangJ Y, et al. Deep-reinforcement-learning-based distributed computation offloading in vehicular edge computing networks[J]. IEEE Internet of Things Journal, 2023, 10(14): 12416-12433.
GuoXiao-dong, HaoSi-da, WangLi-fang. Task offloading method based on deep reinforcement learning for vehicular edge computing[J]. Application Research of Computers, 2023, 40(9): 2803-2807, 2814.
KuangZhu-fang, ChenQing-li, LiLi-feng, et al. Multi-user edge computing task offloading scheduling and resource allocation based on deep reinforcement learning[J]. Chinese Journal of Computers, 2022, 45(4): 812-824.
XuXiao-long, FangZi-jie, QiLian-yong, 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.