Study of some practical issues in inversion with the Born iterative method using time-domain data

The performance of the Born iterative method of nonlinear two-dimensional profile inversion is examined for the reconstruction of large objects and in the presence of measurement noise. Time-domain data are used. It is shown that objects at least as large as about nine wavelengths can be inverted without any convergence problems. The algorithm is shown to perform well in the presence of 10% noise, or 20-dB signal-to-noise ratio. The simultaneous inversion of permittivity and conductivity profiles is formulated and solved using the Born iterative method. Objects with various loss tangents are reconstructed, and the limits of applicability of the algorithm are investigated.<
>