Shape Reconstruction from Experimental {Radar Cross Section, Phase} Data (contributed talk)

The inverse electromagnetic problem for a perfectly conducting obstacle is stated in two spatial dimensions. Part One of the talk deals with the heuristics of shape reconstruction. Two classes of algorithms are presented, ABP and AFP. The former, known as the approximate back propagation algorithm, minimises the boundary defect; the latter, known as the approximate forward propagation algorithm, minimises the far-zone defect. Numerical results are provided for both. Part Two attempts at justifying the algorithms. The least-squares boundary coefficients are investigated in relation to error bounds and the Rayleigh hypothesis. The affine-least squares scheme is investigated in relation to the approximate forward propagator. The convergence of said propagator holds, provided the spectral radius of a Gram matrix is less than one.