ML-Based Detection, DoA Estimation and Localization of MultiRotor and Fixed-Wing UAVs Over Restricted Areas in Cluttered Environment

Project: Research

Project Details


Drones and unmanned arial vehicles (UAVs) are becoming ubiquitous with many potential civilian and environmental applications. They have been deployed in battlefields for surveillance and/or attacks among other military purposes [1]. Such prevalence and capabilities raised concerns of threats if their payloads carry explosives and have been controlled/programmed to breach into highly restricted areas as in critical oil and gas facilities. The existing counter unmanned aerial systems rely mainly on radar, electro- optical/infrared (EO/IR), RF directional finding (RFDF), and acoustic sensors. The challenge is that drones and UAVs are different in size, endurance/flight time, payload weight, and flying aerodynamics, e.g. multi-rotor or fixed-wing, causing various Doppler signatures in received radar signals/echoes. They are contaminated by background clutter which has non-Gaussian and slowly varying signature. Due to their size and material, they have a low radar cross section (RCS). Moreover, those traveling at low speed and altitude will result in dense clutter and multipath effect -especially if the facility to be protected is in an urban area [2]. In this project, we will develop novel detection algorithms to identify unauthorized aerial vehicle(s) under strong cluttered environment using machine learning. This will require an accurate statistical characterization to model both the target(s) and channel (urban/rural), and we will optimize the radar parameters to enhance the probability of detection and therefore its accuracy. It is expected that the results will have a direct impact in the development and enhancement of anti-drone systems and to our national security
Effective start/end date1/07/2131/12/22


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.