无线供能及移动边缘计算技术的整合为下一代无线通信网的实现提供了技术支持.然而,用户数量的激增将对诸如系统响应时效性和超低延时等需求的实现提出了新的挑战.因此,如何设计迭代次数少、收敛速度快、灵活性强的实时计算卸载策略成了研究的新热点.文章梳理了无线供能移动边缘计算(Wireless Powered Mobile Edge Computing,WP-MEC)系统在实现超低延时需求上面临的问题与挑战;总结了WP-MEC系统的网络模型及其计算卸载策略的研究概况;详细阐述了 4 种不同接入方式下的WP-MEC系统的计算卸载策略研究现状;对比分析了各类传统的数值优化方法及深度强化学习优化方法在实时计算卸载决策方面的优劣;对低复杂度高效计算卸载策略的发展进行总结与展望,提出了延时最小化计算卸载策略的3 个关键研究方向.
Review of research on offloading strategy for computing delay optimization in wireless powered edge computing
The integration of wireless energy supply and mobile edge computing technology provides a technical support for the realization of the next generation wireless communication network.However,the surge in the number of users poses new challenges to the realization of requirements such as system response timeliness and ultra-low latency.Therefore,how to design a real-time computing offloading strategy with fewer iterations,fast convergence and strong flexibility has become a new research focus.This paper combs the problems and challenges faced by wireless powered mobile edge computing(WP-MEC)system in achieving ultra-low latency requirements,summarizes the research overview of WP-MEC system network models and computing offload strategies,and describes in detail the research status of computing offloading strategies of WP-MEC systems under four different access modes.This paper compares and analyzes the advantages and disadvantages of various traditional numerical optimization methods and deep reinforcement learning optimization methods in real-time computing offloading decision-making.This paper summarizes and prospects the development of low complexity and efficient computing offloading strategies,and proposes three key research directions of delay minimization computing offloading policy.
deep reinforcement learningwireless powered networkmobile edge computingoffloading strategycomputing delay minimization
万方数据
维普
