Fingerprinting IoT Devices using Artificial Intelligence Techniques

Alex Hoskins, Ashley Renwick, Stephen Hopkins, Dr. Caroline John and Dr. Ezhil Kalaimannan, Department of Computer Science, Hal Marcus College of Science and Engineering, University of West Florida, Pensacola, FL 32514

Internet of things (IoT) devices are becoming increasingly prominent in home and business networks. Developing a way of fingerprinting IoT devices is necessary as these devices pose big security risks unsupervised. Therefore, a need exists to be able to identify and track such devices on a network in order to manage allowed and possibly vulnerable devices, and rogue IoT devices that could serve as an attack vector. In this research project, we address these IoT security issues with a focus on their application software. The breaches are very hard to predict: even when using data prediction models based on collected network traffic information from both rogue and authorized devices. Prior neural networking research established a level of detection over 96% as an operating system and 95% as an independent machine. 

Our goal is to extend the neural network for operating system detection towards an application for IoT devices to achieve passive detection of rogue IoT systems as a group and individually. We also want to classify their characteristics and behavior on a network. There will be five areas of classification: Trusted and Normal Operations Behavior, Asset Ownership, Correlation of Access Control Credentials to IP Addresses, and Network Activity in Absence of Owners. We will capture live data in order to analyze and classify it with respect to the defined patterns of behavior using fastai deep learning libraries. The tools and methodology we develop should be able to efficiently detect rogue devices on a network. We will also be able to analyze and classify the IoT devices on a network based on our research findings.


Keywords: Internet of Things (IoT), Operating System Fingerprinting, Neural Networks, Artificial Intelligence, and fastai.



Additional Abstract Information

Presenters: Alex Hoskins, Ashley Renwick

Institution: University of West Florida

Type: Oral

Subject: Computer Science

Status: Approved

Time and Location

Session: Oral 3
Date/Time: Mon 4:30pm-5:30pm
Session Number: 310
List other presenters in this same room and session