Surface Reconstruction of Large Reflector Antennas Based on a Hybrid of CMA-ES and HIO Algorithms

Abstract

By taking advantages of the global search ability of the covariance matrix adaptation evolution strategy (CMA-ES) and the quick local convergence rate of the hybrid input-output (HIO) algorithm, this paper constructs a practical phase retrieval algorithm used in the surface reconstruction for large reflector antennas. The proposed technique, named CMAES-HIO, is a hybrid two-steps global optimization algorithm which only requires one far-field intensity pattern as the input. Numerical simulation results show that the proposed algorithm is effective to avoid the local stagnation of the conventional iterative phase retrieval algorithms.