Parallel extended-born analysis of electromagnetic scattering from 3-Dimensional sub-rough surface targets

Tiejun Yu, Lawrence Carin

Research output: Contribution to journalArticlepeer-review

Abstract

Real 3-dimensional rough surface (RS) half space is treated as a perfect half-space with a special target which leads to the huge problem size if integral equation (IE) based method of moments (MoM) is used, so parallel algorithm is applied to speedup the EM scattering calculation from the RS. Message-passing interface (MPI) is the most widely used new standard for parallel calculation. It is not a new programming language, rather it is a library of subprograms that can be called from C(++) and Fortran programs. In the paper, a powerful cluster with 33-CPU units connected by MPI network technique provides a good chance to investigate the real 3D RS half space EM scattering. Instead of rigorous MoM, a more efficient algorithm named extended-Born (E-Born) is used to model this 3D RS scattering problem. As E-born is a very natural parallel algorithms, the complete parallel E-born code is almost P times fast as a series E-born code, while P is the CPU number in the cluster. With N is the problem size which is usually very large for real 3D-RS scattering, the LU-decomposition (LUD) solution of MoM is of order N3 complexity, while a series E-Born solution is of order 27N, and the parallel E-Born solution is at the order of 27N/P. With such a computation complexity of 27N/P, a problem with size N > 105 is no longer a prohibitive task. The accuracy and efficiency of the parallel E-Born method are validated by the MoM results.
Original languageEnglish (US)
Pages (from-to)260-263
Number of pages4
JournalIEEE Antennas and Propagation Society, AP-S International Symposium (Digest)
Volume4
DOIs
StatePublished - Jan 1 2002
Externally publishedYes

Fingerprint

Dive into the research topics of 'Parallel extended-born analysis of electromagnetic scattering from 3-Dimensional sub-rough surface targets'. Together they form a unique fingerprint.

Cite this