Small Unmanned Aerial Systems Detection and Classification Using IEEE 802.11 Signals
Open Access
- Author:
- Emani, Rahul
- Graduate Program:
- Cybersecurity Analytics and Operations
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- March 28, 2022
- Committee Members:
- Edward Glantz, Professor in Charge/Director of Graduate Studies
Nick Giacobe, Thesis Advisor/Co-Advisor
Anna Cinzia Squicciarini, Committee Member
Edward J Glantz, Committee Member
Timothy E. Kohler, Special Signatory - Keywords:
- sUAS
UAVs
Cybersecurity
Drones
IEEE 802.11
Signals
Network Traffic Analysis
Drone Detection
Drone Classification - Abstract:
- Small Unmanned Aerial Systems (sUAS) or drones are enjoying great popularity due to their ease of use, low cost, and wide availability. Today, these systems can be piloted using cellular devices as well as other networked systems. Given their size, sUAS are generally undetectable, can be launched from almost anywhere, and have little to no regulations regarding takeoff and flying. Depending on the airframe design and application, a given system may even have the ability to carry a sizeable payload, e.g., agricultural drones used to spray crops. Given the aforementioned statements, one can readily see why these same “hobby systems” are also very popular with those wishing to invade privacy or cause harm and damage. For example, nothing is preventing someone with bad intentions from filling the spray reservoir with a fluid containing viruses or bacteria to launch a bioterrorism attack. Along the same line, if the airframe is of sufficient size to carry a container of liquid, there is nothing preventing someone from replacing the container with an explosive device. Therefore, there is a strong need to know when any drone is operating within a given geographic dome. This thesis will thus explore the feasibility of using IEEE 802.11 signals for the detection and classification of Wi-Fi enabled hobbyist drones to provide drone operation accountability. In the end, drone flight data was collected, and analyzed for differentiating features to aid in drone detection and classification. Then, seven different machine learning algorithms were trained and evaluated for their effectiveness of differentiating a drone versus a wireless access point. After that, the seven machine learning algorithms were retrained and reevaluated for their effectiveness at identifying the specific make and model of drone. Together, the research conducted explored a drone detection and classification framework desperately needed for invoking C-sUAS responses.