Traffic Fingerprinting Attacks on Internet of Things Using Machine Learning

The Internet of Things (IoT) concept has been widely adopted and Internet connected devices enter more and more areas of our everyday lives. However, their limited security measures raise increasing concerns, especially in terms of users' privacy. That is why, in this paper, privacy risks, focusing primarily on information leakage exposed by traffic fingerprinting attacks, on IoT devices are investigated. The considered attacks take advantage of the statistical network flows' features and application of machine learning (ML) methods and can be utilized by a passive traffic observer. In this perspective, the first part of the research presented in this paper analyzes the feasibility of identifying individual devices in a victim's home network. It considers smart environment setups of different scales and conditions, and it also includes a performance comparison of the different ML models applied. The second part introduces and validates a method for the devices' state detection based on pattern recognition with ML. Finally, recommendations for mitigating the discussed privacy risks are also enclosed.