An Effective Computing Cloud–Edge-Seamless System for Industrial Internet of Things
提出云边无缝系统CESS，融合云计算和边缘计算，通过OBRI和NOB方案减少延迟、平衡负载，适用于工业物联网中复杂网络场景。
Growth of computing capacity at edge drives edge computing as an efficient computing model, and cloud computing is considered as an indispensable computing model as various cloud services emerge. However, edge computing performs gracelessly in persistent tasks and the tasks requiring dispersed data, whereas cloud computing is often accompanied with high latency. To deal with more complex network scenarios, we fuse the two computing models and propose cloud-edge-seamless system (CESS) which enables all infrastructures to orchestrate data as services in a 5G/6G network, realizing infrastructure as a service (IaaS). In CESS, raw data is processed on intermediate processing devices (IPDs) before reaches the cloud. Thereafter, we proposed OBRI scheme to minimize processing and routing delay in which IPD can orchestrate data as a service if it is busy and handle data roughly if it is idle, thus reducing data volumes and total delay according to IPD congestion. Furthermore, there is still a large number of flows queuing on downstream IPDs, then NOB scheme is proposed to reduce delay further, in which IPD sends notice to upstream IPDs when congestion occurs to restrict them to orchestrate data only. Thus, the flow velocity slows down and more flows are orchestrated as services upstream, resulting greater delay reduction. Extensive experimental results show that the proposed CESS is immensely efficient. OBRI and NOB can effectively reduce delay, balance load, and improve quality of experience (QoE). In addition, NOB provides great flexibility and is extremely suitable to process delay-sensitive data.