GraphSAGE-Enhanced Security Authentication Protocol for Internet of Things(loT) Devices

Abstract

A new model for device authentication in large-scale Internet of Things (IoT) systems has been proposed based on graph representation learning methods. In heterogeneous and resource-constrained IoT environments, the efficiency, flexibility, and security of authentication protocols are the three main topics investigated in this paper. Using GraphSAGE with context-aware feature aggregation and neighborhood encoding, dynamic device interactions are modeled as evolving graphs. The purpose of the ablation study is to build a full-climate experimental system to simulate and benchmark adversarial attacks on multiple networks. The results show that under minimal pressure, the success rate of impersonation attacks on this protocol is very low, at only 1%, and it consistently blocks over 97% of privilege escalation attempts. Authentication delay is stable; as the number of devices increases from 100 to 20,000, it increases sub-linearly, with the median delay rising from only 120 milliseconds to 310 milliseconds. Compared to the previous lightweight protocol, this protocol reduces both communication and computational costs by over 30%. Through ablation experiments, key design elements such as the context aggregation module have been shown to be crucial for system stability and error suppression. In summary, this paper provides an excellent high-scalability solution for trusted IoT authentication, offering quantitative security metrics and performance data across various deployment environments.