A Method for Anomaly Detection of Edge Time Series Data and Network Programmable Control

Abstract

The invention relates to an abnormality detection and network programmable control method of edge time series data, and belongs to the field of combining time series data of the Internet of Things with deep learning and machine learning. By obtaining the time series data on the edge devices of the Internet of Things; predicting the time series data of the Internet of Things based on the attention mechanism based on Grid LSTM; predicting the time series data of the Internet of Things on the edge devices through the prediction model of the attention mechanism based on Grid LSTM to get the real The error between the value and the predicted value; the SVM algorithm is used to detect the abnormality of the above error, and the abnormal situation of the data is obtained; the traceability and shielding of the transmission path of the abnormal data packet are realized, and the search for the new transmission path of the data is realized. The invention has the beneficial effects of improving the ability of analyzing and processing time-series data of the Internet of Things, improving data prediction performance and anomaly detection performance; and solving the data security problem during wireless sensor network data transmission.

Description

An edge time series data anomaly detection and network programmable control method technical field The present invention relates to a kind of edge sequential data abnormality detection and network programmable control method, belong to Internet of Things sequential The field of combining data with deep learning and machine learning. Background technique [0002] With the development of smart city, industry 4.0, supply chain and home automation technology, the Internet of Things (Internet of Things, IoT) applications have generated a large amount of data. According to Cisco report, by 2021, connected IoT devices The number will reach 11.6 billion, meaning more than 49 exabytes of data traffic will be generated each month. Ubiquitous sensors generate A large amount of data and information, and these data are becoming the most common form of data in IoT computing, making data transmission And processing plays an increasingly critical role in IoT applications. Through...

AI Technical Summary

Problems solved by technology

Although there are many application scenarios for current time series data and spatiotemporal data prediction in real life, most of them are used for energy-saving analysis of nodes in wireless sensor networks, and these methods are rarely used to consider the security of wireless sensor nodes

Images