Machine learning methods for signal classification in anti-drone technology

Drone technology has recently gained wide-scale acceptance in multiple military and commercial applications, including security surveillance and medicine delivery. In 2017 alone, the market revenue from drones is estimated to be around $6 billion. With this, there is also an emerging need for sophisticated anti-drone technology to detect rogue drones entering secure territories, such as military bases and prisons. In this project, we will design machine learning algorithms which can detect unknown drone radio signals from a mixture containing radio signals transmitted by multiple other wireless devices, including WiFi and Bluetooth. We will analyze several datasets of radio signals transmitted by popular drones in the market, so as to extract unique fingerprints hidden in the drone signals. Machine learning models will then be designed to make use of the extracted fingerprints for detecting the presence of a drone signal when a mixture of radio signals is fed as input.

Intern: 
Naga Raghavendra Surya Vara Prasad Koppisetti
Faculty Supervisor: 
Vijay Bhargava
Province: 
British Columbia
Partner University: 
Program: