Spectral acceleration of the generalized forward-backward method

The generalized forward-backward (GFB) method was introduced as an iterative method of moments solution to compute the electromagnetic scattering from targets on ocean-like rough surfaces. In this paper, an existing spectral acceleration algorithm is adapted to the GFB. The spectral acceleration is presented for both polarizations and for perfect electrically conducting (PEC) and non-PEC surfaces. The accelerated scheme reduces the computational scaling cost of the GFB from O(N/sup 2/) to O(N) per iteration as the length of the sea surface increases linearly with the number of unknowns N. The numerical results demonstrate that the spectral acceleration introduces negligible error while greatly improving the efficiency of the GFB method.