Date: Friday, July 17th, 2023
Time: 10 AM – 12 PM ET
Conference Room: Klaus Conference Room 2100
Virtual Location: Click here to join the meeting
Alan Nussbaum
Ph.D. Candidate
School of Computer Science
College of Computing
Georgia Institute of Technology
Committee:
Dr. Umakishore Ramachandran (Advisor) - School of Computer Science, Georgia Institute of Technology
Dr. Dale Blair (Co-Advisor) - School of Electrical and Computer Engineering, Georgia Institute of Technology and Georgia Tech Research Institute
Dr. Vivek Sarta - School of Computer Science, Georgia Institute of Technology
Dr. Ada Gavrilovska - School of Computer Science, Georgia Institute of Technology
Dr. Tushar Krishna - School of Electrical and Computer Engineering and School of Computer Science, Georgia Institute of Technology
----------------------------------------
Abstract:
Traditional radar systems are often susceptible to high-maneuverability targets capable of sudden changes in velocity and acceleration. These changes in a target’s trajectory degrade the signal-to-noise ratio (SNR) in the Signal Processor, thereby limiting the track precision of the radar and blurring a target's range and angle measurements, an effect called range walk. While the radar is tracking the target's kinematic state (position, velocity, and acceleration), an optimal signal processing algorithm requires knowledge of the target's range rate and radial acceleration to reprocess the sensor-received data. Different research approaches have been proposed to solve this problem with limited success and without adaptation for multiple potential scenarios and architectures to achieve low latency and stringent sensor timelines. This dissertation takes advantage of an iterative Expectation-Maximization algorithm (EM-based), which executes enhanced range walk compensation algorithms in a real-time control loop software architecture that models and prototypes computational and program steering of the sensor software components and computer resources. The EM-based algorithm uses existing signal processing and tracking algorithms with functional, logic, and execution modifications that are implemented in the real-time software architecture (Hypersonic). This Hypersonic architecture is a nontraditional sensor architecture that modifies the traditional pipeline by providing unknown data (velocity and acceleration) to the Signal Processor for reprocessing of the radar sensor data to create a higher fidelity measurement for the track filter (i.e., mitigate SNR loss). This dissertation has implemented and demonstrated an integrated set of EM-based signal processing and tracking algorithms that can achieve real-time performance resulting from the steering software architecture that implements task and data parallelism. The real-time software architecture adheres to latency and timeline requirements that are critical for a deterministic sensor architecture and result in an optimization of critical radar sensor resources.
]]>