Robustness of Sparse Reconstruction Methods in Spherical Antenna Near-Field Measurements

The robustness of reconstruction strategies for the application of compressed sensing to a spherical antenna nearfield to far-field transformation is investigated. More precisely, the minimum number of measurement samples is determined for different noise levels such that a predefined far-field accuracy is achieved. Considered are the strategies iterative hard thresholding, hard thresholding pursuit, orthogonal matching pursuit, and compressive sampling matching pursuit in a modified form, as well as the quadratically constrained basis pursuit (BP). Modified phase transition diagrams show that depending on the noise and sparsity level a different recovery strategy should be used, while in general the quadratically constrained BP is outperformed by the other considered strategies.