Content area
Full Text
(ProQuest: ... denotes non-US-ASCII text omitted.)
Research Papers
I.
INTRODUCTION
Within the security and defense domain, radar is more and more applied in the confined and crowded urban and littoral environments. Consequently, there is a demand for detecting and classifying a wider range of small targets such as mopeds, dismounts, animals, birds, flocks of birds, and mini-drones. Basically, detection of these smaller targets requires lowering the detection threshold, with respect to both target radar cross section (RCS) and Doppler velocity. However, the sheer number of objects in crowded littoral and urban environments may potentially saturate the radar signal processing, leading to, e.g. lost tracks. Ultimately, situational awareness is affected.
In these environments, full situational awareness can be maintained only if target classification can be done reliably and rapidly. Rapid classification allows filtering-out objects that are irrelevant for the current mission. For this first rapid classification, distinction between broad target classes may be sufficient. Depending on the mission, these broad classes could be man-made object, i.e. a potential threat, and bio-life, i.e. a non-threat, such as a bird. In a next classification step, it is desired to provide further separation within the potential threat classes to aid the threat assessment. For instance, the size or number of rotors of a drone may be an indication of its maximum payload. In this paper, the potential of exploiting micro-Doppler properties for this two-step classification approach will be reviewed.
The classification problem addressed within the current study focuses on recognizing small unmanned aerial vehicles (UAVs). Mini-UAVs are an emerging threat and exhibit so-called "LSS" characteristics, for Low (altitude), Small (RCS), and Slow (speed), which makes them challenging radar targets when they operate in an environment with for instance birds.
To reduce the number of false alarms, it is important to quickly classify a UAV as a man-made object, preferable before the tracking stage where the identity of all objects currently present is maintained, and thus the number of irrelevant objects should be minimal to prevent track overload. In the next step, further characterization of the UAV is desired. Some characteristics of interest are the type of UAV, the number of rotors, approximate size, etc. This classification can be done by a trained human operator just by...