A BANDED MATRIX ITERATIVE APPROACH TO MONTE-CARLO SIMULATIONS OF LARGE-SCALE RANDOM ROUGH-SURFACE SCATTERING - TE CASE

A banded matrix iterative approach is applied to study scattering of a TE incident wave from a perfectly conducting one-dimensional random r ough surface. It is much faster than the full matrix inversion approac h or the conjugate gradient method. When compared to the Kirchhoff ite rative approach, it is of comparable CPU time, and works for cases whe n the Kirchhoff iteration is erroneous. The method is illustrated for a variety of parameters with particular application to large scale rou gh surface problems. The largest surface length used is 400 wavelength s with 3200 unknowns, and all the coherent wave interactions axe inclu ded within the entire surface length. The accuracy of the banded itera tive approach is demonstrated by showing that the results overlie thos e of the exact matrix inversion and the conjugate gradient method. The numerical method is also easy to implement. With this approach, we ar e able to compute the new response characteristics of composite rough surfaces with much larger scales. The case of large incidence angle is also studied. Comparison is also made with the analytic second-order Kirchhoff theory.