重庆邮电大学学报（自然科学版）2025，Vol.37Issue(2)：222-231,10.DOI:10.3979/j.issn.1673-825X.202403150055
数字孪生辅助MEC任务卸载和资源分配联合优化
Digital twin-assisted joint optimization of MEC task offloading and resource allocation
摘要
Abstract
In densely deployed mobile edge computing(MEC)scenarios,the heterogeneity of edge server resources and the dynamic complexity of environmental conditions pose significant challenges for designing efficient task offloading and re-source allocation strategies.To address this issue,a digital twin edge network(DTEN)model is established.With the goal of minimizing long-term task processing latency under multi-dimensional constraints on bandwidth,computation,and stor-age resources,a digital twin(DT)-assisted joint optimization problem is formulated for MEC user association,task offload-ing,service caching,and resource allocation.Given the highly dynamic nature of the network,the optimization problem is modeled as a partially observable Markov decision process(POMDP),and a deep reinforcement learning(DRL)-based task offloading scheme is designed.The DRL model is trained using mapping data from physical entities within the DT.Fur-thermore,considering the delay deviation between DT estimations and actual values,the impact of this deviation on task processing latency is analyzed.Extensive simulation experiments demonstrate that the proposed scheme effectively reduces task processing latency and improves cache hit rates.关键词
移动边缘计算/任务卸载/资源分配/数字孪生/深度强化学习Key words
mobile edge computing/task offloading/resource allocation/digital twins/deep reinforcement learning分类
信息技术与安全科学引用本文复制引用
李云,江源,王碧,梁吉申,夏士超..数字孪生辅助MEC任务卸载和资源分配联合优化[J].重庆邮电大学学报（自然科学版）,2025,37(2):222-231,10.基金项目
国家自然科学基金项目(62301099,62071077) （62301099,62071077）
中国博士后科学基金项目(2023MD734137) （2023MD734137）
重庆市自然科学基金项目(2022NSCQ-LZX0191) （2022NSCQ-LZX0191）
重庆市教委科学技术研究项目(KJQN202300638) National Natural Science Foundation of China(62301099,62071077) （KJQN202300638）
China Postdoctoral Science Foundation(2023MD734137) （2023MD734137）
Natural Science Foundation of Chongqing(2022NSCQ-LZX0191) （2022NSCQ-LZX0191）
Science and Technology Research Program of Chongqing Municipal Education Commission(KJQN202300638) （KJQN202300638）
