Constantly improving ladar sensor technology has pushed simulation capabilities required for hardware-in-the-loop sensor testing and algorithm development beyond the capabilities of standard desktop PCs. Robust ladar computations require transport and manipulation of large, complex, multi-dimensional datasets containing range, irradiance, micro-Doppler, polarization, speckle decorrelation, and other information. Coherent Technologies, Inc. (CTI) is developing a portable, scalable software architecture for implementing ladar imaging simulation calculations on large cluster-based supercomputers. This architecture takes advantage of both line-of-sight and transverse modes of parallelization for the various stages of computation encountered in typical ladar calculations. In order to assure portability of software, this effort has followed ANSI coding standards for C/C++ and parallel data control is implemented using the Message Passing Interface (MPI). Using this rather general coding framework, CTI researchers have realized parallel efficiencies in excess of 50%, or fixed problem speedups of up to 19x on 32 processors. As increased fidelity is incorporated into the simulator, parallel efficiency is expected to improve even further.

Date Published: 19 May 2005
PDF: 12 pages
Proc. SPIE 5791, Laser Radar Technology and Applications X, (19 May 2005); doi: 10.1117/12.609693

Published in SPIE Proceedings Vol. 5791:
Laser Radar Technology and Applications X
Gary W. Kamerman, Editor(s)