Docker-Enabled Scalable Parallel Mlfma System for RCS Evaluation

Abstract

Abstract-Research on RCS evaluation for electrically large objects has been a hot topic for decades. Although multilevel fast multipole algorithm (MLFMA) has been the most popular method in scattering computation, due to the limitation of both CPU speed and memory size in a single computer, realistic large targets require discretization with millions of unknowns still cannot be solved by sequential implementations of MLFMA. In this paper, we introduce a Docker-enabled parallel MLFMA computing system based on MPI, which is proved to be friendly for deployment and economical for scalability, to solve electrically large scattering problems. In addition, the capability of the proposed system has been demonstrated by several canonical examples.