能量采集型边缘计算架构下的节能任务卸载与高能效资源分配方法研究

Edge computing, as a new computational model that can perform calculations at the edge of the network, can provide low-latency and high-bandwidth network services. In response to the national call for energy conservation and environmental protection, combined with energy harvesting technology, the edge computing can also provide renewable green energy for the user side and the edge side, and reduce the use of non-renewable energy in the network. This project focuses on task offloading decision and resource allocation strategy in the edge computing with energy harvesting. For the problem of unstable energy harvesting, it plans to use Virtual Machine (VM) migration technology and wireless power transfer technology to study how to achieve the goal of energy-saving task offloading and high-energy-efficient resource allocation in a scenario with sufficient energy and limited energy. The specific research contents include: (1) The VM migration mechanism and task offloading strategy for dynamic random queues will be studied, and the online Lyapunov method will be adopted to simplify the global dynamic optimization model as the static optimization model of decision variables. To achieve a balance between optimal target and system stability, the VM migration and task offloading decisions for each time slot will be optimized. (2) Studying benefit-driven based resource allocation method, designing auction and matching algorithm, establishing dynamic resource pricing strategy, realizing rapid resource allocation and maximizing the system’s energy efficiency.

为响应国家节能环保的号召，边缘计算作为一种在网络边缘执行计算的新型计算模型，可结合能量采集技术，在提供低延时、高带宽网络服务的同时为用户侧和边缘侧提供可再生的绿色能源，减少网络对非再生能源的使用。本项目专注于能量采集型边缘计算架构下的任务卸载决策和资源分配策略，针对能量采集不稳定的问题，拟采用虚拟机迁移技术和无线能量传输技术，研究如何在能量充足和能量有限的场景中，实现节能任务卸载和高能效资源分配的目标。具体的研究内容包括：（1）研究面向动态随机队列的虚拟机迁移机制和任务卸载策略，拟使用在线Lyapunov方法化简全局动态优化模型为决策变量的静态优化模型，优化每一个时隙的虚拟机迁移和任务卸载决策,取得目标最优和系统稳定间的平衡。（2）研究基于效益驱动的资源分配方法，拟设计拍卖算法和匹配算法，建立动态资源定价策略，实现资源快速分配，最大化系统能量效益。

本项目主要研究了边缘计算中的任务卸载和资源分配优化技术，先后提出了一种边缘计算中的请求分配和计算资源分配联合优化方法、一种多层计算网络中的收入最大化服务功能链的在线部署方法和一种面向边缘计算的可靠性和延迟约束的能耗最小化任务卸载方法等。在理论上和实际实验上，上述方法都被证明了其优异的性能，相比其他最新的方法和成果，在能耗、延迟、计算速度等方面都有显著的提升。项目执行过程中先后获得授权发明专利6项，并在《IEEE Transactions on Services Computing》、《IEEE Transactions on Parallel and Distributed Systems》等国际学术期刊和国际会议上发表论文共计7篇。此外，在人才培养方面，已培养毕业硕士研究生5名，在培养硕士研究生6名。项目经费开支合理。