Research on Physical Layer Authentication Method of Internet of Things Based on Contour Stellar Images and Convolutional Neural Network

Wireless communication networks are prone to information security issues due to the openness of radio transmission. User information security relies heavily on reliable identification and authentication methods. To address the channel differences between different users, this paper proposes a physical layer identity authentication method using contour stellar images. By converting one-dimensional signals into two-dimensional contour stellar images using measured channel information from two channel scenarios, and enhancing the data set with data enhancement technology, a convolutional neural network is employed to classify and recognize 800 contour stellar images in two scenarios. This achieves the goal of different user identity authentication. Additionally, the CNN model and SVM model are compared under the same conditions. The results indicate that the proposed CNN model achieves 98.3% recognition accuracy, demonstrating strong robustness and authentication efficacy in real-world scenarios.