AbstractA novel, low-cost, phased-array antenna that uniquely incorporates bulk phase shifting using voltage-tunable dielectric (VTD) material is presented. The array does not contain an individual phase shifter at each radiating element. This paper presents the antenna concept and describes how it can be used as a low-cost phased array. The VTDs that are used in this antenna are described. The measured antenna patterns of a prototype phased array demonstrating electronic beam scanning at 10 GHz are also presented.
Abstract
Achieving robust and fast two-dimensional adaptive beamforming of phased array antennas is a challenging problem due to its high-computational complexity. To address this problem, a deep-learning-based beamforming method is presented in this paper. In particular, the optimum weight vector is computed by modeling the problem as a convolutional neural network (CNN), which is trained with I/O pairs obtained from the optimum Wiener solution. In order to exhibit the robustness of the new technique, it is applied on an 8 × 8 phased array antenna and compared with a shallow (non-deep) neural network namely, radial basis function neural network. The results reveal that the CNN leads to nearly optimal Wiener weights even in the presence of array imperfections.
Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas