Date of Award
2015
Document Type
Thesis
Degree Name
Master of Science in Engineering (MSE)
Department
Electrical and Computer Engineering
Committee Chair
Laurie L. Joiner
Committee Member
Mark Tillman
Committee Member
W. David Pan
Subject(s)
deciBel Research, Ultra-wideband devices, Radar--Automatic detection, Adaptive signal processing, Radar targets--Computer simulation, Target acquisition--Computer simulation
Abstract
Effective, radar-based, missile defense requires an efficient, accurate determination of whether an object is lethal or non-lethal. An ideal method would accurately classify targets that contain unknown variations with a minimal number of radar pulses. A wideband, single pulse, manifold classifier is evaluated. The algorithm, deciBel Research's Target Attribute Surface Manifold (dBTASM), is tasked with correctly classifying the pieces of a ballistic missile complex. For this experiment, the algorithm has a database of only three objects, each representing a different piece of the complex. Against this database, missile pieces of different sizes and configurations were classified. An effort to improve classification results through the use of different distance metrics was made. These metrics characterize the fit of the return pulse to the database, and thus they affect the robustness of the algorithm to object variations. Results were mixed; no distance metric proved clearly superior. Recommendations for future work are presented.
Recommended Citation
Weir, Laura Hicks, "dBTASM(TM) robustness study" (2015). Theses. 136.
https://louis.uah.edu/uah-theses/136