Air Surveillance Radar Classification Improvement

Many fielded radar systems, including legacy systems often throw useful classification information on the floor, resulting in less-than-optimal performance in identifying and distinguishing objects of interest from other objects and clutter. There are several reasons this may occur: (1) the classifiers do not adequately exploit the physics of the radar waveforms interacting with the objects in track; (2) the classifiers use legacy techniques that do not take advantage of recent developments in machine learning (ML) algorithms to combine all available information; or (3) the classifiers do not adequately exploit the time-series information, which can capture discriminating object motion patterns and features. To properly classify Unmanned Aerial Systems (UAS), and distinguish them from other objects including manned aircraft and vehicles, ground clutter, birds, etc., nou Systems, Inc. (nSI) is proposing to develop an advanced track classification algorithm suite for Counter Target Acquisition (CTA) radars, applicable to the AN/TPQ-53 and AN/TPQ-50, to discriminate targets of interest from other objects including clutter, and reduce false alarms from benign objects. nSI has the subject matter expertise, experience, tools and techniques ready to meet the Army need, and will leverage many ground-breaking techniques recently developed and demonstrated for other DoD applications, including USAF legacy radar systems for Missile Defense missions. These techniques focus on exploiting the physics-based features unique to the specific object classes and radar frequencies, and fusing these features using industry-proven ML methods.