ECE Seminar Speaker: Dr. Dayalan Kasilingam, ECEDate(s): 4/19/2013 1:00 PM - 4/19/20132:30 PM
Location: Lester W. Cory Conference Room, SENG - Room 213A
Contact: Honggang Wang firstname.lastname@example.org 508-999-8469
Topic: “Reducing the Computational Complexity of Problems in Applied Electromagnetics”
In applied electromagnetics, one is interested in solving problems related to radiation, propagation and scattering of electromagnetic waves such as radio waves (RF), microwaves and light. In scattering problems, the scattering of electromagnetic signals such as wireless communications signals and radar signals by objects and surfaces are analyzed. When analyzing a scattering problem, one starts with solutions predicted by Maxwell’s equations and then arrives at a specific solution by enforcing the boundary conditions for a particular object or surface. Generally the solution involves two types of waves – propagating waves and non-propagating waves known as evanescent waves. Fields associated with evanescent waves are storage fields such as those in capacitors and inductors. In many practical applications, one is not interested in knowing the strength of the evanescent fields. In radar applications, the strength of the radar echo is determined by the scattered propagating fields and not by the evanescent fields. However, when one solves a scattering problem by enforcing boundary conditions, one has to compute the strength of both the propagating fields and the evanescent fields. In this study, we have developed an innovative technique by which we are able to compute the strengths of the propagating fields without computing the strengths of the evanescent modes. This is accomplished by identifying a sub-space in the signal space which is orthogonal to the evanescent modes. The boundary conditions are then projected on to this sub-space and the solutions are computed by solving for only the propagating fields. In this presentation, we show the existence of a unique signal sub-space which is orthogonal to the evanescent fields by analyzing the scattering of waves from periodic surfaces. We show that the results obtained by this method are consistent with the full-blown solution which includes evanescent fields. We conclude by showing a novel technique for estimating this signal sub-space and how this technique helps to reduce the computational complexity of scattering problems.
Dr. Dayalan Kasilingam is Chair and Professor of Electrical and Computer Engineering at UMD. His research interests are in the areas of radar remote sensing and wireless communications. He currently works on problems in electromagnetic scattering, super-resolution imaging and on signal modulation techniques for high-data rate, mobile wireless communications. Dr. Kasilingam’s research has been supported by grants from NSF, ONR and NASA. In 1995, he became UMD’s first recipient of the prestigious NSF CAREER award. He also led a team which won the Silicon Graphics Supercomputing contest by applying the simulated annealing technique to extract geophysical information from remotely sensed satellite data.