J.E. Bendz, H.G. Fernandes, and M.K. Zuffo (Brazil)
Finite-difference time-domain (FDTD) methods, higher or der schemes, numerical methods, simulation optimization.
Higher order finite-difference time-domain schemes are generally used to either improve the accuracy of the nu merical solution of Maxwell’s equations, or take advantage of the reduced truncation errors to be able to use a coarser grid discretization of the electrical domain. In this paper we propose an algorithm to optimize the normally fixed co efficients used in the four-point central-differences for the standard higher order (2,4) finite-difference time-domain algorithm. This allows us to use a coarse discretization of the three-dimensional environment, that reduces the high demand for memory (> 23 times) and the overall computa tional runtime (> 28 times) compared to the standard finite difference time-domain method, and still produces very ac curate results for less dense discretizations. A theoretical analysis of the expected very low numerical dispersion er rors has been carried out, as well as numerical simulations to confirm these values. The usefulness of the model under realistic circumstances has also been validated by perform ing a large-scale three-dimensional simulation of a typical Wi-Fi environment.
Important Links:
Go Back
