Analysis of Next Generation Sensor Data
Dr. Dean Keiswetter | Acorn SI
SERDP has overseen development of a new generation of digital electromagnetic induction (EMI) sensors that provide significant advantages to unexploded ordnance (UXO)/clutter discrimination. The new sensors include the Naval Research Laboratory (NRL) time-domain electromagnetic (TEM) array, the Berkeley UXO Detector (BUD), and the Advanced Ordnance Locator (AOL2), and they differ from previous designs (e.g., EM63, GEM3) in several ways, including the absence of navigation errors due to rigid multicoil construction and the presence of bistatic interrogation (multiple transmit/receive pairs). These systems offer the prospect that previously difficult problems of data inversion may be addressed successfully.
The objective of this project is to enhance the capabilities of these new sensors by developing methods to exploit the advantages they offer in three specific areas: (1) multiple target environments, (2) nondipole effects, and (3) parameter search issues. These parallel efforts are intended to have a complementary effect, conferring an overall improvement greater than the sum of the components.
This project will develop, implement, and test innovative solutions for the three areas mentioned above. For multiple target environments, three complementary approaches will be explored—one based on sparse matrices, the second based on the Method of Auxiliary Sources (MAS), and the third based on the Discrete Dipole Approximation (DDA). For nondipole effects, researchers will implement and evaluate fast indexing of precomputed corrections to the dipole model. For parameter search issues, they will systematically characterize various proven techniques as applied to the new sensor systems.
While this research will be tailored to the TEM as a test bed, results also will be applicable to the BUD, the AOL2, and to any system in general that employs rigid multicoil construction and bistatic interrogation. Reliable multitarget inversion will lead to cost savings by reducing the zone at the center of a target bulls-eye, which must be entirely excavated. Addressing nondipole effects also will lead to cost savings through better accuracy in recovered target response values (betas), producing better UXO/clutter discrimination. The benefit of improved parameter search methods is faster and more robust data inversion.