Intelligent Detection Method for Malicious Attacks in Enterprise Internet of Things Based on Graph Neural Network

With the continuous development of Internet of Things (IoT) technology and the widespread application of IoT by enterprises, enterprise IoT is facing increasingly complex and frequent malicious attack threats. To effectively detect malicious attacks in enterprise IoT and timely take corresponding security protection measures, an intelligent detection method was proposed for malicious attacks in enterprise IoT based on graph neural networks. Pre-processing enterprise IoT data, extracting data features, analyzing data packet transmission characteristics, and employing a graph neural network to extract time series features were involved in this method. To mitigate feature gradient disappearance, additional connections were incorporated. Data containing interactive nodes was filtered using these features, and a three-layer model was established for classifying the data, thus facilitating intelligent detection of malicious attacks. Experimental results show that this method can accurately describe the changing characteristics of malicious attacks on enterprise IoT and overcome the shortcomings of current malicious attacks on enterprise IoT. It achieves a 100% accuracy rate on the KDD (Knowledge Discovery and Data Mining) Cup 1999 dataset and enterprise IoT and is consistent with the imported data information. The average malicious node detection rate of this method is 97.26% on the two datasets. The method outperforms node-centered control algorithms, mutual information-based methods, and GWB-LSSVM joint methods. The proposed intelligent detection method exhibits strong feasibility, high accuracy, and important practical application value. © 2023 School of Science, IHU. All rights reserved.