工业智能边缘计算中基于数据流相关性的计算卸载研究

Research on Computing Offloading Based on data Stream Correlation in Industrial Intelligent Edge Computing

大数据时代背景下为满足工业物联网的高实时性需求,我们运用边缘计算和机器学习技术提出了一种工业智能边缘计算中基于数据流相关性的计算卸载方法。采用拓扑排序、决策树等方式寻求任务相关性,避免无关数据卸载、多任务场景下重复数据多次卸载及无序卸载导致的缓存过多,存储器容量不足等问题。本研究设计了两种调度算法,分别对应多任务决策所需特征数据重复率高和低的两种不同场景下所采用的计算卸载算法,保证了任务处理的实时性;此外本研究均严格控制在带宽约束下,保证了调度的可靠性。

In order to meet the high real-time demand under the background of industrial Internet of Things in the era of big data, we propose a task offloading method based on data stream correlation in industrial intelligent edge computing based on edge computing and machine learning. Topological sorting, decision tree are adopted to seek task relevance, so as to avoid problems such as excessive cache and insufficient memory capacity caused by irrelevant data offloading, repeat data offloading multiple times and unordered offloading in multi-task scenarios. In this study, two scheduling algorithms are designed, respectively corresponding to two different scenarios of high and low repetition rate of characteristic data required by multi-task decision so as to ensure the real-time performance of task processing. In addition, this study is strictly controlled under the constraint of bandwidth to ensure the reliability of scheduling.