A lightweight internet of things intrusion detection method, system, device and storage medium

Abstract

The application discloses a lightweight Internet of Things intrusion detection method, system, device and storage medium, and belongs to the technical field of Internet of Things intrusion detection, and the method comprises the following steps: performing feature screening preprocessing on an original Internet of Things traffic data set; synthesizing a few-class attack sample by adopting a denoising diffusion probability model without a classifier guide to obtain an enhanced data set; after the enhanced data set is randomly divided, the enhanced data set is input into an MBConv-Transformer hybrid detection architecture for training, the architecture extracts local features through an MBConv-ECA module, models global dependence through a Transformer module, and realizes multi-scale feature fusion; and after pretreatment of to-be-detected data, the to-be-detected data are input into the trained model to output an intrusion detection result. The application solves the problem that in the Internet of Things intrusion detection, network traffic data classes are extremely unbalanced, a few-class attack detection false negative rate is high, and existing deep learning models have large parameters and are computationally intensive, and are difficult to adapt to the lightweight deployment demand of Internet of Things terminal resources.

Images