Millimeter wave hybrid beam forming design method based on deep reinforcement learning

Abstract

The invention discloses a millimeter wave hybrid beam forming design method based on deep reinforcement learning, and the method comprises the steps that a base station obtains an analog precoding matrix of a user through the calculation of cross-correlation according to the obtained channel state information of the user; constructing a deep reinforcement learning intelligent agent for jointly optimizing the digital precoding matrix and the analog merging matrix of the user, inputting user channel information and the analog precoding matrix into the intelligent agent, and outputting the corresponding digital precoding matrix and the analog merging matrix; and calculating a digital merging matrix of the user based on a minimum mean square error criterion. According to the invention, the millimeter wave hybrid beam forming design method based on deep reinforcement learning is high in convergence speed and good in robustness, and the spectral efficiency of the system can be effectively improved.

Description

technical field [0001] The invention relates to a millimeter-wave hybrid beamforming design method based on deep reinforcement learning, and belongs to the technical field of adaptive transmission of a point-to-point MIMO downlink system configured with a uniform linear antenna array at a base station. Background technique [0002] As an effective method that can increase network transmission rate and alleviate the shortage of spectrum resources, millimeter wave communication is regarded as one of the key technologies of the new generation wireless communication network. The millimeter-wave communication system combined with massive multiple-input multiple-output (MIMO) can make full use of space resources, realize multiple transmission and multiple reception through multiple antennas, and double the system channel capacity without increasing spectrum resources and antenna transmission power , and at the same time effectively solve the problems of weak penetrating power and ...

AI Technical Summary

Problems solved by technology

Although the time complexity is reduced after applying these methods, certain system performance is sacrificed

The deep supervised learning method has high requirements on the amount of training data and is very sensitive to channel fading changes. For millimeter-wave channels with weak penetration and fast attenuation, it is still difficult to apply in practice.

Images