5G Radar and Wi-Fi Based Machine Learning on Drone Detection and Localization
Mark Ignatius Teo, Chee Kiat Seow, Kai Wen
Abstract
Drone usages have been proliferating for various government initiatives, commercial benefits and civilian leisure purposes. Drone mismanagement especially civilian usage drones can easily expose threat and vulnerability of the Government Public Key Infrastructures (PKI) that hold crucial operations, affecting the survival and economic of the country. As such, detection and location identification of these drones are crucial immediately prior to their payload action. Existing drone detection solutions are bulky, expensive and hard to setup in real time. With the advent of 5G and Internet of Things (IoT), this paper proposes a cost effective bistatic radar solution that leverages on 5G cellular spectrum to detect the presence and localize the drone. Coupled with K-Nearest Neighbours (KNN) Machine Learning (ML) algorithm, the features of Non- Line of Sight (NLOS) transmissions by 5G radar and Received Signal Strength Indicator (RSSI) emitted by drone are used to predict the location of the drone. The proposed 5G radar solution can detect the presence of a drone in both outdoor and indoor environment with accuracy of 100%. Furthermore, it can localize the drone with an accuracy of up to 75%. These results have shown that a cost effective radar machine learning system, operating on the 5G cellular network spectrum can be developed to successfully identify and locate a drone in real-time.