Trustworthiness Evaluation System of UEIOT Devices Based on Deep Learning

• Published in: 2023 International Conference on Networks, Communications and Intelligent Computing (NCIC) pp. 116 - 122
• Main Authors: Chen, Cen, Wang, Ming, Li, Nuannuan, Lv, Zhuo, Li, Mingyan, Chang, Hao
• Format: Conference Proceeding
• Language: English
• Published: IEEE 17.11.2023
• Subjects: CNN; Computational modeling; Data mining; Deep learning; DQN; Feature extraction; Internet of Things; MLP; Smart grids; Training; UEIOT
• DOI: 10.1109/NCIC61838.2023.00026

With the rising popularity of smart devices and the growth of IoT applications, AI technology is thriving in edge computing environments. These environments provide ample opportunities for applying AI due to the abundance of realtime sensing data and diverse application scenarios. Moreover, AI technology facilitates the extraction of value from edge data, thus driving the development of emerging information industries, particularly in areas like smart grids. However, the security of IoT devices in smart grids faces significant challenges, making it difficult to swiftly and effectively assess the trustworthiness of such devices. This poses implications for ensuring information security in IoT and edge computing. This paper introduces existing models for evaluating IoT devices, identifies shortcomings in current systems, and proposes a trust evaluation method for UEIOT devices based on deep learning. The experimental results demonstrate the following contributions of this paper: (1) Transformation of CNN's traditional map feature extraction into the extraction of IoT device connection data for effectively evaluating IoT devices over a historical period. (2) Application of the DQN model analogy to evaluate ubiquitous power IoT devices and simulate interactions between intelligent body devices and external IoT devices.