Millimeter wave MIMO hybrid beam forming optimization method based on deep learning

Abstract

The invention discloses a millimeter wave MIMO hybrid beam forming optimization method based on deep learning. According to the millimeter wave MIMO hybrid beam forming optimization method based on deep learning, constraint conditions in a traditional millimeter wave large-scale MIMO hybrid beam forming optimization problem can be mapped into a neural network, and a multi-user hybrid beam formingsystem is completely converted into an equivalent neural network. Therefore, a complex non-convex optimization problem in hybrid beam forming can be converted into end-to-end unsupervised optimizationsimilar to an auto-encoder, and joint optimization can be carried out on a plurality of beam forming matrixes.

Description

Technical field [0001] The invention relates to the field of wireless communication, in particular to a millimeter wave MIMO hybrid beamforming optimization method based on deep learning Background technique [0002] Hybrid beamforming technology is a promising technology for millimeter wave multiple input multiple output (MIMO) systems to support ultra-high transmission capacity and low complexity. However, the design of digital and analog beamformers is a challenge with non-convexity optimization, especially in multi-user situations. The optimization problem of hybrid beamforming involves the optimization of four beamforming. The use of a layered structure to gradually optimize the four matrices cannot ensure the global optimal solution; the beamforming matrix obtained by the researcher proposed by the traditional scheme is used as the label training neural network. Method, its performance is restricted by traditional schemes, and it does not make full use of the powerful appr...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to overcome the above-mentioned deficiencies in the prior art, and to provide an optimization method for millimeter-wave MIMO hybrid beamforming based on deep learning. The present invention uses a neural network to construct a hybrid beamforming system, and the hybrid beam The matrix F of the baseband beamformer in the shaping system bb , the matrix F of the RF beamformer rf , the matrix W of the RF combiner rf and baseband combiner matrix W bb Equivalently converted into a cascaded neural network comprising four neural networks; in the present invention, the radio frequency combined neural network is split into k sub-radio frequency combined neural networks; the baseband combined neural network is split into k sub-baseband combined neural networks; the sub-radio frequency combined neural network It is connected with the sub-baseband combined with the neural network one by one to simulate the multi-user scenario in a single cell; and the user priority coefficient is added to change the user optimization priority; the entire beamforming system is mapped to a neural network, so that the mixed beamforming The complex non-convex optimization problem is transformed into an end-to-end unsupervised optimization technique similar to autoencoders to solve the joint optimization of beamforming matrices in hybrid beamforming techniques

Images