面向边缘计算的异构资源协同与全局效能优化研究

With the advancement of Internet-of-Things, billions of end devices are expected to be connect to the internet and generate explosive amounts of IoT data. However, due to the long communication latency between end-devices and cloud datacenters, cloud computing is unable to satisfy the stringent timeliness requirement of processing the huge amount of IoT data in a real-time manner. To address this challenge, the concept of edge computing has been recently proposed to process the IoT data at the network edge, and thus to significantly reduce the communication latency. Unfortunately, compared to the cloud, the edge is less energy-efficient, computing capable and cost-efficient. Based on the intuition of coordinating heterogeneous resources to exploit diverse advantages, in this project we propose to scale the edge resource from three different directions, and thus to mitigate the issues of energy-inefficiency, cost-inefficiency and limited computing capacity of the edge. Specifically, by scaling-down to coordinate the end-devices, we exploit the idle resource of huge amounts of end-devices to support device-to-device cooperative computing, and thus to reduce the energy consumption of the edge node. By scaling-up to coordinate the cloud, we carefully distribute the workload between the edge and cloud based on the heterogeneity of resource price, and thus to reduce the operational cost. Finally, by scaling-out to coordinate other edge nodes, we incentivize multiple edges to share the computing resources among edges, and thus to improve the resource utilization.

随着物联网技术的蓬勃发展，云计算技术已经无法满足大规模终端设备对海量数据实时处理的迫切需求。针对这一难题，边缘计算这一新型计算模式应运而生。相比于云计算节点，边缘计算节点虽在实时性上具有明显优势，但在计算能力、能源效率和成本效率这三个方面却存在显著劣势。因此，研究边缘计算的效能优化问题对于节约成本和实现可持续发展的绿色边缘计算至关重要。基于协同异构资源从而实现差异性优势互补这一基本思想，本项目拟从三个不同的方向实现异构资源协同，从而解决边缘计算节点所面临计算能力低、能源效率低和成本效率低的问题。向下协同终端，整合海量终端设备空闲资源来支持设备间协同计算，从而降低边缘节点能耗；向上协同云端，基于资源价格差异性优化计算负载在边缘和云端之间的分配，从而降低计算成本；向外协同其它边缘节点，激励邻近边缘节点实现跨边缘的资源共享，从而突破单个边缘节点资源限制并提升整体资源利用率。

随着物联网技术的蓬勃发展，云计算技术已经无法满足大规模终端设备对海量数据实时处理的迫切需求。针对这一难题，边缘计算这一新型计算模式应运而生。相比于云计算节点，边缘计算节点虽在实时性上具有明显优势，但在计算能力、能源效率和成本效率这三个方面却存在显著劣势。因此，研究边缘计算的效能优化问题对于节约成本和实现可持续发展的绿色边缘计算至关重要。基于协同异构资源从而实现差异性优势互补这一基本思想，本项目拟从三个不同的方向实现异构资源协同，从而解决边缘计算节点所面临计算能力低、能源效率低和成本效率低的问题。向下协同终端，整合海量终端设备空闲资源来支持设备间协同计算，从而降低边缘节点能耗；向上协同云端，基于资源价格差异性优化计算负载在边缘和云端之间的分配，从而降低计算成本；向外协同其它边缘节点，协同邻近边缘节点实现跨边缘的资源共享，从而突破单个边缘节点资源限制并提升整体资源利用率。