Toward Secured Internet of Things (IoT) Networks: A New Machine Learning based Technique for Fingerprinting of Radio Devices

The process of identifying and authenticating Internet of Things (IoT) devices based on the electromagnetic characteristics of their wireless interfaces is a topic that has been receiving a lot of attention lately due to the continuous growth in the market of embedded and wearable wireless devices. Securing the networks to whom these devices are connecting daily has become a significant security challenge, which is threatening the security and safety of thousands, maybe millions of private and public networks due to the vulnerable nature of wireless devices to a well-known set of possible attacks.In this paper, we present the initial results acquired from our work-in-progress to develop new radio-features-extraction-based technique to identify wireless devices, specifically Internet of Things (IoT) ZigBee and LoRa devices. This paper summarizes our initial experimental setup to gather and analyze the devices signals to extract the desired features and describes the signal pre-processing approach and the machine learning methods that we attempted to date.