time-modulated antenna array

In this article, the synthesis of the two-dimensional radiation pattern of multi-beams in the time-modulated planar antenna array is discussed. With the aim of reducing the time of numerical calculations of multi-beam synthesis and eliminating the traditional and optimization approaches, which are mostly complicated and time-consuming, the convolutional neural network approach has been investigated. In this study, the simultaneous shaping of multiple beams as desired in the time modulated antenna array is presented for the first time. By using the method of switching elements and their time modulation, which is based on the Chabi-Sheff distribution, to use and realize multi-beams such as the fundamental beam and the first and second harmonics with low side lobe level and steering at different spatial angles, create various random data. After that patterns and modulations like them are stored. After that, by presenting and designing a model of the convolutional neural network, learning the model for the relationship between the main beam pattern and the first two harmonic patterns with the time modulation of each element has been done. The presented neural network has been able to learn the relationship between the time modulation parameters of the antenna array elements with the main beam patterns and the first and second harmonics with a mean square error of about 0.03. In order to evaluate this model, a random sample of data has been selected to give its patterns to the input of the network. The output of the network has estimated the time modulation sequence of each element. Finally, the modulation pattern of the estimated elements is compared with the main patterns of the comparison and shows the closeness of the original pattern to the estimated pattern. This method provides a good capability for arbitrary control of multiple beams for applications that require establishing multiple connections.