Protecting Smart Homes: A study on AI approaches to Detect Malwares and Developing a Two Factor Authentication System Based on SRAM PUF For Smart Home Users in UAE

Abstract

The UAE vision towards sustainability is increasingly promoting the adoption of smart home technology for its residents and is expected to grow over 10% on a yearly basis until 2029, while smart technology offers convenience and efficiency to smart home inhabitants. However, at the same time smart home users are vulnerable to cyber incidents if these devices are not secured against the cyber-attacks. Due to the rapid adoption of smart technologies in smart homes, the Open Web Application Security Project has identified weak or easily guessable passwords as a significant vulnerability in IoT devices, based on real-world incidents. The current research seizes this research gap as an opportunity to address unauthorized access threats in smart homes by proposing a dual approach: one component emphasizes malware detection via machine learning, while the other incorporates a two-factor authentication system which integrates a hardware-based security feature, the static random-access memory (SRAM) Physical Unclonable Function (PUF), with a user-created password, providing enhanced protection against unauthorized access. Numerous existing methods, such as multi-factor authentication, frequently proved inadequate in practical scenarios, failing to resolve problems like weak passwords and insecure connections. The current research consists of two phases. During Phase 1, the current research employs DC and Random Forest models to detect malware in network traffic, with an exceptional accuracy of 99.99% and an AUC-ROC score above 0.9999. Whereas in Phase 2, the SRAM PUF was implemented and validated in controlled laboratory settings. The experimental findings of the current research report successfully mitigation of unauthorized access through internal and external networks thus enhancing the user’s security in smart homes environment. Although the proposed approaches in current research, i.e., Machine Learning and SRAM PUF function independently, machine learning concentrates on malware detection, while SRAM PUF guarantees device authenticity, they together mitigate any un-authorized access. The findings of the current research were analyzed in a controlled environment to comply with ethical considerations and results were validated using thematic analysis. Thus, enhancing the security of smart home users.