Antenna array radiation pattern efficient small-batch synthesis method based on deep learning network in 5G application field

Abstract

The invention provides an antenna array radiation pattern efficient small-batch synthesis method based on a deep learning network in the 5G application field. The method comprises the following steps of: 1, designing a target data matrix to describe important characteristics of different synthetic target radiation directional diagrams according to requirements of a target directional diagram; and 2, determining the structure and parameters of the deep network according to the characteristics of different network layer structures, and taking the difference between a target data matrix and a radiation pattern predicted by a neural network as a loss function in training to guide the gradient descent direction. And 3, describing all target radiation directional diagrams needing to be synthesized as a target data matrix, combining the target data matrix into a tensor, inputting the tensor into a neural network, carrying out small-batch parallel computing, and completely outputting the amplitude and phase information of the antenna array units meeting the corresponding target design requirements at one time.

Description

technical field [0001] The invention relates to the technical field of deep learning and antenna array radiation pattern synthesis, in particular to an efficient small batch synthesis method for antenna array radiation patterns based on a deep learning network in 5G applications Background technique [0002] The wide application of 5G technology promotes the development of information technology and continues to affect other technical fields. Compared with 4G technology, 5G can provide larger communication capacity, higher communication speed and lower communication delay, which makes it popular in industrial Internet, Internet of Vehicles, medical and other fields. Millimeter wave communication technology is one of the main means of 5G communication technology. However, due to the poor penetration and short wavelength of millimeter waves, the communication quality is easily disturbed by environmental factors. In view of the above problems, multiple antenna units are arran...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problems in the synthesis of antenna array radiation patterns in the field of 5G applications, as the total number of different radiation patterns simultaneously calculated increases, the time overhead increases and the synthesis effect is not ideal. This paper proposes a method based on deep learning. An Efficient Small Batch Synthesis Method for Antenna Array Radiation Patterns for Networks

Images