Inversion algorithm for optimizing multi-antenna systems

In this paper, we present the results of a method for optimizing multi-antenna systems from imposed constraints (e.g. ideal radiation patterns at a single frequency or over a wide frequency range). The method is based on a surface integral formulation of the forward EM problem and a gradient method for minimizing a cost functional. The originality of the method lies in the analytic calculation of derivative of the cost functional versus different parameters such as location and shape. Various configurations of systems involving planar antennas illuminated by dipoles are studied. The robustness of the algorithm is investigated with noise-corrupted data.