Affiliation:
1. School of Information Science and Technology, North China University of Technology, Beijing 100144, China
2. Department of Computer Science, University of Illinois Springfield, Springfield, IL 62703, USA
Abstract
With the rapid expansion and ubiquitous presence of the Internet of Things (IoT), the proliferation of IoT devices has reached unprecedented levels, heightening concerns about IoT security. Intrusion detection based on deep learning has become a crucial approach for safeguarding IoT ecosystems. However, challenges remain in IoT intrusion detection research, including inadequate feature representation at the classifier level and poor correlation among extracted traffic features, leading to diminished classification accuracy. To address these issues, we propose a novel transformer-based IoT intrusion detection model, MBConv-ViT (MobileNet Convolution and Vision Transformer), which enhances the correlation of extracted features by fusing local and global features. By leveraging the high correlation of traffic flow, our model can identify subtle differences in IoT traffic flow, thereby achieving precise classification of attack traffic. Experiments based on the open datasets TON-IoT and Bot-IoT demonstrate that the accuracy of the MBConv-ViT model, respectively, 97.14% and 99.99%, is more effective than several existing typical models.
Funder
National Natural Science Foundation of China
Reference30 articles.
1. Intrusion Detection System for Large-Scale IoT NetFlow Networks Using Machine Learning with Modified Arithmetic Optimization Algorithm;Fraihat;Internet Things,2023
2. (2024, May 01). The Growth in Connected IoT Devices Is Expected to Generate 79.4zb of Data in 2025, According to a New IDC Forecast. Available online: https://www.businesswire.com/news/home/20190618005012/en/The-Growth-in-Connected-IoT-Devicesis-Expected-to-Generate-79.4ZB-of-Data-in-2025-According-to-a-New-IDC-Forecast.
3. Chollet, F. (2017, January 21–26). Xception: Deep Learning with Depthwise Separable Convolutions. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA.
4. ResNet-50 vs. EfficientNet-B0: Multi-Centric Classification of Various Lung Abnormalities Using Deep Learning “Session Id: ICMLDsE.004”;Kansal;Procedia Comput. Sci.,2024
5. Chhabra, M., and Kumar, R. (2022). A Smart Healthcare System Based on Classifier DenseNet 121 Model to Detect Multiple Diseases. Mobile Radio Communications and 5G Networks: Proceedings of Second MRCN 2021, Springer Nature.