Signature Prediction and Uncertainty Analysis for Radar-based MDA Applications

Period of Performance: 02/17/2011 - 02/16/2013


Phase 2 SBIR

Recipient Firm

Jin Consulting, Inc.
2808 Willow Bend Rd.
Champaign, IL 61822
Principal Investigator


The objective of this proposal is to develop user-friendly, highly accurate and efficient, robust and powerful computer codes for predication of radar signatures of MDA objects of interest (MOIs) and the effects of geometrical and material uncertainties. Three computer codes will be developed, one for body-of-revolution (BOR) targets, another for discrete body-of-revolution (DBOR) targets, and the third for scattering by a deep, arbitrarily-shaped air duct in an air-breathing missile. Each code will be self-complete as it will include a user-friendly graphical user interface (GUI), a solid modeling capability that can accept commonly used geometry data formats, a robust meshing capability to generate high-quality meshes, a parallelized main computational engine, and necessary post-processing capabilities to handle output data. Equally important, each code will have unmatched capabilities to model complex geometries and materials that can be inhomogeneous and anisotropic. All the three codes are based on a novel physics-based computational electromagnetics (CEM) algorithm for solving Maxwell s equations and a robust stochastic collocation-based algorithm for uncertainty analysis, both of which have been studied and fully validated in the feasibility study conducted in Phase I. The CEM algorithm combines the finite element and boundary integral methods and implements a novel numerical technique to exploit either the continuous or the discrete rotational symmetry of a BOR or DBOR target to accelerate computations. For scattering by a deep cavity --a well-known grand challenge in CEM, the hybrid finite element and boundary integral method employs a special frontal algorithm to speed up calculation and reduce memory requirements. The stochastic collocation algorithm is capable to perform an uncertainty analysis by using a significantly smaller number of samples than required by the traditional Monte-Carlo method. The development of the proposed computer codes will provide a highly accurate and efficient and easy-to-use tool to compute the radar signatures of MOIs and to quantify potential errors due to geometrical/material modeling uncertainty. All the codes will be parallelized to harness the power of parallel computing. The use of graphical processing units (GPU) will be explored to further speed up the computations.