ACES Publication Search

This paper presents the results to date of an on-going development effort to provide a scalable, portable, parallel scene generation tool that will provide the capability to rapidly generate scenes of radiating and scattering structures in realistically complex electromagnetic environments. The benefit of such a tool is that it will provide users with the capability to solve large problems that cannot be currently solved with existing sequential electromagnetic modeling tools, This tool supports a broad range of users including researchers, algorithm developers, analysts, and system developers. This paper will present the parallelization process highlighting the strategies used and will show the results to date. The project presented here is the parallelization of WIPL-D, an electromagnetic modeling tool. Through parallelization, the well known and commercially available tool will become faster and possess increased capabilities. This paper will walk you through the parallelization process, providing the strategies used and the results received.

This paper presents the results of the final year of a development effort to provide a scalable, portable, parallel scene generation tool that provides the capability to rapidly generate scenes of radiating and scattering structures in realistically complex electromagnetic environments. The benefit of such a tool is that it will provide users with the capability to solve large problems that cannot be currently solved with existing sequential electromagnetic modeling tools. This tool supports a broad range of users including researchers, algorithm developers, analysts, and system developers. This paper will present the parallelization process and will show the final results of the project. The project presented here is the parallelization of WIPL-D, an electromagnetic modeling tool, which picks up in time from where [1] left off. Through parallelization, the well known and commercially available tool became faster and now possesses increased capabilities. This paper will walk you through the parallelization process, providing the strategies used and the results received.

WIPL-D (WIre PLate Dielectric) has become an increasingly popular Method of Moments (MoM) code used in computational electromagnetics (CEM) modeling. WIPL-D was chosen for parallelization under the Common High Performance Computing Software Support Initiative (CHSSI) program of the High Performance Computing Modernization Office (HPCMO). Hence, the new code was given the name WIPL-DP, where \"P\" stands for Parallelized. Any computer code chosen for parallelization under the CHSSI program must undergo four rigorous phases of testing: Software Acceptance Test, Alpha Test (AT), Beta Test (BT), and Initial Operational Test and Evaluation (IOT&E). WIPL-DP is currently undergoing those tests, with the Alpha Test recently completed and reported on in this effort. The Beta Test and the IOT&E are to be completed by 30 September 2004. WIPL-DP is a parallelized C/C++ version of the original FORTRAN 77 WIPL-D code. During the Alpha Test period, WIPL-DP was successfully parallelized for frequency. It also received optimal performance rating for the following Critical Technical Parameters (CTPs): scalability; portability; and correctness, stability, and accuracy. The chosen test case for the Alpha Test was a modified version of the “Human Head Adjacent to a Cellular Phone” (DEMO-531) problem available under the tutorial sub-directory in the PC version of the WIPL-D software. The Alpha Test was performed on two distinct High Performance Computing (HPC) platforms, Tempest and Huinalu, both at the Maui HPC Center.

WIPL-D is being used to compute monostatic and bistatic radar cross sections of a trihedral corner reflector over the frequency range of 1-12 GHz. Initial results are discussed for vertically and horizontally polarized fields at two frequencies (1.5 and 3.8 GHz). These computations are being used to provide a benchmark against which the performance of WIPL-DP, a parallelized version of the WIPL-D, may be compared. Recently, sea-scatter data was collected from 1.9-11.5 GHz with trihedral reflectors. Since the WIPL-D calculations discussed here, as well as planned calculations, will be used to provide predictions of sea scatter that will be compared to the actual data, the sea-scatter geometry is described.

Electromagnetic analysis of large complex structures and structures in the environment requires massive processing memory and computational capability. The ever-increasing speed and memory size of today\'s computers is dwarfed by the need to solve complex problems. There exists a highly efficient commercial electromagnetic modeling tool called WIPL-D that partially addresses these problems. This paper will present a discussion of an effort that is currently underway to develop a parallelized WIPL-D that will expand the problem sizes that can currently be solved. The parallelized WIPL-D will provide exciting new design and research possibilities for electromagnetic analysis.