| 摘要: |
| 在移动边缘计算网络中，联合考虑用户的移动性和服务满意度，实现数字孪生（Digital Twin，DT）的有效部署是一个极大的挑战。针对该问题，提出了一种边缘网络中基于用户移动性预测的数字孪生部署策略。首先，利用双向长短期记忆网络（Bidirectional Long Short-Term Memory Network，Bi-LSTM）和注意力机制建立一种轨迹预测模型，对移动用户进行轨迹序列预测；然后，利用整数线性规划（Integer Linear Programming，ILP）对数字孪生的部署问题进行建模；最后，以最大化由信息新鲜度（Age of Information，AoI）定义的效用增益函数为优化目标，提出一种基于用户移动性预测的数字孪生部署策略算法对提出的问题进行求解。该算法根据获取到的移动用户轨迹数据，利用数字孪生部署的边际效用递减特性进行设计，以实现最优的数字孪生部署策略。仿真分析验证了所提算法在预测精度和效用增益方面的有效性，且该算法与基准算法相比显示出性能提升不低于10.7%。 |
| 关键词: 移动边缘计算 数字孪生 用户移动性预测 部署策略 |
| DOI：10.20079/j.issn.1001-893x.240423002 |
|
| 基金项目: |
|
| Digital Twin Deployment Strategy Based on User Mobility Prediction in Edge Networks |
| ZHOU Dongyang,PAN Xianbing |
| (1.School of Digital Economy and Information Management,Chongqing College of Mobile Communication,Chongqing 401520,China;2.Chongqing Key Laboratory of Public Big Data Security Technology,Chongqing 401420,China) |
| Abstract: |
| In the dynamic networks of mobile edge computing,it is a great challenge for the effective deployment of digital twins(DTs) while jointly considering user mobility and service satisfaction.The authors propose a DT deployment strategy based on the prediction of user mobility in mobile edge networks.Firstly,a trajectory prediction model is constructed by utilizing Bidirectional Long Short-Term Memory Networks(Bi-LSTMs) combined with attention mechanisms to forecast the movement sequences of mobile users.Secondly,the deployment problem of DT is formulated as Integer Linear Programming(ILP) model.Finally,a digital twin deployment strategy algorithm based on user mobility prediction is proposed to solve the problem aiming at maximizing the utility gain function defined by age of information(AoI).The algorithm is designed based on the obtained mobile user trajectory data and utilizes the diminishing marginal utility of digital twin deployment to achieve the optimal digital twin deployment strategy.The simulation results demonstrate that the effectiveness of the proposed algorithm in terms of prediction accuracy and utility gain,and the algorithm demonstrates a performance improvement of no less than 10.7% compared with the benchmark algorithm. |
| Key words: mobile edge computing digital twin user mobility prediction deployment strategy |
《电讯技术》编辑部 