Abstract:

Mobile Edge Computing（MEC） is an application extension of cloud computing technology on top of edge infrastructure. An energy-efficient UAV mobile edge computing system is considered to minimize system energy consumption by jointly optimizing the UAV’s motion trajectory, computation offloading, and resource allocation. In order to solve the above problems, a two-layer optimization method is proposed, which uses real-time clustering of users based on unsupervised learning based channel gain-Self-Organizing Feature Map（h-SOM）, which is based on channel gain as an indicator of the judgment category and get the optimal deployment position of the drone. in the lower layer, according to the deployment of the UAV, the computation offloading and resource allocation problem is converted into a Mixed Integer Nonlinear Programming Problem（MINLP）, and the double-variation Improved Differential Evolution（IDE） algorithm with elite initial strategy and adaptive operator is adopted. The elite initial strategy can improve the accuracy of the algorithm, the double mutation method can improve the global search ability of the algorithm and promote the convergence of the algorithm, thus obtaining better task unloading and resource allocation decision. The effectiveness of the proposed method is verified by simulation experiments, and compared with the traditional algorithm, the optimization effect is significant, which provides a new idea for the joint optimization of MEC system.

Key words: mobile edge computing, h-SOM neural network, Improved Differential Evolution（IDE） algorithm, double mutation strategy

摘要：

移动边缘计算（MEC）是云计算技术在边缘基础设施之上的应用拓展。考虑一个高能效的无人机移动边缘计算系统，通过联合优化无人机的运动轨迹、任务卸载策略和计算资源分配来最小化系统的能耗。为解决以上问题，提出一种双层优化方法，在上层用基于无监督学习的信道增益-自组织特征映射网络（h-SOM）对用户进行实时聚类，该聚类是以信道增益作为判断类别的指标并得到无人机的最佳部署位置；在下层根据无人机的部署，将计算卸载和计算资源分配问题转化为混合整数非线性规划问题（MINLP），并采用带有精英初始策略和自适应双变异策略的改进差分进化算法（IDE）进行迭代求解，精英初始策略可以根据h-SOM的聚类结果提供优秀的初始解，自适应双变异策略能够提高算法的全局搜索能力并促进算法收敛，从而获得更好的任务卸载决策。通过仿真实验验证了所提方法的有效性，并与传统算法进行了比较，其优化效果显著，为MEC系统的联合优化提供了一种新思路。

关键词: 移动边缘计算, h-SOM神经网络, 改进差分进化（IDE）算法, 双变异策略