Distributed Internet of Things equipment anomaly detection method

Abstract

The invention relates to a distributed Internet of Things equipment anomaly detection method, and belongs to the field of Internet of Things. According to the method, potential feature distribution of abnormal data is learned, feature correction and feature discrimination are carried out, high-quality reconstructed data are obtained, the proportion of normal data to abnormal data is balanced, training of an abnormal detection network is guided through the balanced data, a discrimination threshold is output, and the abnormal data are identified. Then, calculating the confidence coefficient of each model by adopting a federal learning algorithm based on dynamic model selection, and dynamically selecting a local model and uploading the local model to a central server for model aggregation; according to the invention, high-precision anomaly detection of high-dimensional and unbalanced data in an Internet of Things scene can be realized, and the generalization ability of the model is guaranteed.

Description

technical field [0001] The invention belongs to the field of the Internet of Things and relates to a method for detecting abnormalities of distributed Internet of Things devices. Background technique [0002] The Internet of Things refers to the connection of any object with the network through information sensing equipment according to the agreed agreement, and information interaction and communication through the information transmission medium. The purpose of the Internet of Things is to achieve long-term and fast connections regardless of location and time, such as mobile devices in application environments such as smart homes, smart cities, and smart transportation. The advent of the 5G era provides a lot of technical support for the development of the Internet of Things. The formulation of 5G standards can well meet the needs of the Internet of Things, such as network speed, capacity, security, etc., which promotes the development of unmanned driving technology and hel...

AI Technical Summary

Problems solved by technology

Complex technologies such as beamforming, massive MIMO, and dense microcells have been spawned. However, these technologies increase the complexity of the system architecture and make network anomaly detection too complicated, especially to deal with intermittent hardware failures in large networks and Misconfiguration brings the following challenges to anomaly detection: 1) massive unlabeled data warfare; 2) data imbalance; 3) high-dimensional data is heterogeneous

[0004] Existing equipment anomaly detection methods regard the original data as normal data, ignore the randomness of the anomaly, and cannot evaluate the impact of abnormal data on model training, and cannot satisfy the differentiation of the model for heterogeneous equipment detection

Traditional anomaly detection has problems such as low detection rate of unbalanced data and poor model generalization ability. Therefore, a new method for anomaly detection of IoT devices is urgently needed

Images