University Events

Description: TOPIC: Fast Multipole Method for Analyzing Electromagnetic Scattering from Large Objects

LOCATION: Lester W. Cory Conference Room, SENG-213A

ABSTRACT:
Recent advances in computational electromagnetics has provided the means to simulate scattering from large objects and radiation from complex antenna structures. These advances offer the potential to solve extremely complex scattering and radiation problems in electromagnetics. As result of this, there is a need for efficient and versatile numerical techniques and algorithms to perform large-scale computations.
In electromagnetic scattering, many problems can be solved by using boundary-integral methods. For the past several decades, the method of moments (MoM) has been a very popular numerical technique used to solve these integral equations. Unfortunately, due to the large computer storage requirement of this method, only small to medium sized problems can be solved. This is because when the relevant integral equation is discretized using the MoM, it results in a system of dense linear equations. Solving such dense linear equations requires matrix inversions and quickly becomes prohibitively demanding as the size increases. In the past few decades fast iterative methods have been developed for expediting the solution of these linear equations. Different techniques such as the Fast Multipole Method (FMM) have been developed to expedite each iteration step. The efficiency is improved because only a few terms of the multipole expansion are needed to achieve the necessary level of accuracy. Another method used to solve electromagnetic scattering problems is the spatial frequency domain method (SFD). The basic strategy of the SFD method is to take advantage of the convolution structure of the integral equation and to use the FFT to save memory and reduce the number of arithmetic operations. SFD is also able to handle spatial derivatives by using algebraic multiplication. The mathematical structure of the SFD method also lends itself well to FMM-type implementation.
The objective of this dissertation research is to investigate the development of a technique which merges the spatial frequency technique with the fast multipole method, so as to provide significant improvements in computational speed, storage and complexity. The new algorithms will be applied to one-dimensional and two-dimensional canonical scattering and radiation problems. The new algorithms will be compared with existing techniques for speed and storage requirements. The ultimate goal of this project is to apply the new technique to rough surface scattering for determining fully polarmetric scattering signatures. These signatures will help retrieve important geophysical information from remotely sensed radar measurements.

NOTE: All ECE Graduate Students are ENCOURAGED to attend.
All interested parties are invited to attend. Open to the public.

Advisor: Dr. Dayalan Kasilingam

Committee Members: Dr. Antonio H. Costa and Dr. Paul Gendron, Department of Electrical & Computer Engineering; Dr. Alfa Heryudono, Department of Mathematics; and Dr. Branislav Notaros, Electrical & Computer Engineering Department, Colorado State University

*For further information, please contact Dr. Dayalan Kasilingam at 508.999.8534, or by via email at dkasilingam@umassd.edu.