GPU-accelerated spherical near-to-far field transformation software for antenna near-field measurements

Abstract

In the field of antenna measurements, spherical nearfield scanning technique has been well established after years of development. This technique involves the near-to-far field transformation algorithm. The fast development of 5G wireless communication techniques with more applications of the massive wide-band antennas, demands more efficient spherical nearfield measurement systems. Therefore, the speed of the near-to-far field transformation deserves much more attention than before. This paper introduces the GPU-acceleration techniques to the transformation algorithm and develops highly efficient and vectorized Matlab codes. It can accurately treat the electrically large problems (larger than 500 wavelength). Current PC with medium CPU and GPU can treat a 500-wavelength antenna in 90 seconds, which exceeds the claimed speed of the IEEE standard algorithm SNIFT running on a single PC. This paper proves the efficiency of the GPU-acceleration technique in the spherical near-to-far field transformation algorithm.