PowerTrust: AI-Based Trustworthiness Assessment in the Internet of Grid Things

Abstract

Artificial intelligence (AI) is transforming the electrical grid by incorporating advanced communication protocols and novel monitoring infrastructure. This transformation alters traditional methods of electricity consumption and billing, shifting from manual meter reading to dynamic peak-hour tariff rates. While people are open to upgrading their energy consumption systems with cutting-edge technology, they express concerns over transmitting their data over wireless networks, fearing unauthorized access to their private information for unpredictable purposes. Thus, establishing trust in the new technology is crucial before individuals feel secure about sharing their personal information. In this study, we introduce PowerTrust, an ensemble learning stacking model designed to evaluate trust and safeguard user privacy in the Internet of Grid Things. PowerTrust is divided into two parts. In the first part, it assesses the trustworthiness of smart grid devices. In the second part, it proposes a secure scrambling method to protect electricity readings before they are transmitted to the control center. During trust evaluation, data is balanced, important features are extracted, and then a classification model is applied. The synthetic minority oversampling technique is used to balance the dataset, and recursive feature elimination is used to select important features. The results show that the proposed scheme is secure and efficient in maintaining trust and privacy in the grid environment.

Keywords

• AdaBoost-LSTM ensemble learning model
• control center
• home appliances
• PowerTrust
• smart meter
• trust management