基于高空平台的边缘计算卸载:网络、算法和展望

Computation Offloading in HAP-based Edge Computing:Network,Algorithm and Prospect

高空平台(high altitude platform,HAP)技术与多接入边缘计算(multi-access edge computing,MEC)技术的结合将MEC服务器部署区域由地面扩展到空中,打破传统地面MEC网络的局限性,为用户提供无处不在的计算卸载服务。针对基于HAP的MEC卸载研究进行综述,首先,从HAP计算节点的优势、网络组成部分、网络结构、主要挑战及其应对技术4个方面分析基于HAP的MEC网络;其次,分别从图论、博弈论、机器学习、联邦学习等理论的角度对基于HAP的MEC卸载算法进行横向分析和纵向对比;最后,指出基于HAP的MEC卸载技术目前存在的问题,并对该技术的未来研究方向进行展望。

The combination of high altitude platform(HAP)technology and multi-access edge computing(MEC)technology extends the deployment area of MEC servers from the ground to the air,breaks the limitations of traditional terrestrial MEC networks and provides users with ubiquitous computation offloading services.This survey conducts a review on the researches of computation offloading in HAP-based MEC.First,the HAP-based MEC network was analyzed from four aspects:advantages of HAP computing nodes,network components,network structures,main challenges and their coping techniques.Second,from the perspectives of graph theory,game theory,machine learning,federated learning and other theories,the computation offloading algorithms based on HAP were analyzed horizontally and compared vertically.Finally,the current problems of computation offloading technology in HAP-based MEC were pointed out and the future research directions of this technology are prospected.