Planar Near-Field Measurements Based on Kernel Ridge Regression

Due to the development of unmanned aerial vehicle (UAV) hardware and antenna near-field measurement post-processing techniques, the use of UAV to perform nearfield measurements of antennas outdoors is gradually becoming a new situation. Near-field (NF) measurements of the antenna under test (AUT) using an unmanned aerial vehicle-mounted probe and post-processing of the NF data can be used for antenna radiation pattern evaluation and antenna troubleshooting. However, during the actual flight of the UAV outdoors, the measurement trajectory may be deviated by environmental factors such as wind and GPS positioning uncertainty, which may lead to data failure at certain near-field scanning points and affect the post-processing results. This paper focuses on the measurement errors caused by the trajectory offset of UAV and proposes a method of recovering the failed near-field data points based on the kernel ridge regression algorithm to calibrate the errors and reduce the errors of UAV measurements due to external factors. Finally, the simulated near-field sampling is performed, the NF data are sparse processed to simulate the failure data, the recovered complete near-field data are obtained by using the kernel ridge regression algorithm, and the post-processing is performed to obtain the antenna radiation pattern to verify the feasibility.