A Deep Learning-Based MIMO Hybrid Beamforming Method

Abstract

The present invention relates to the field of communication technology, specifically a MIMO hybrid beamforming method based on deep learning, including a channel characteristic information acquisition module, a full digital beamforming simulation module, a sending end matrix module, a receiving end matrix module, and a channel characteristic information acquisition module. Real part and imaginary part module, ResNet neural network model, feature fusion module and training result discrimination module, on the basis of step-by-step optimization and training of the joint channel matrix, the performance of the all-digital beamforming system realized by software simulation is adopted in the present invention As the basis for judging the best performance, the training result is infinitely close to the performance of the full digital beamforming system through the feature fusion of the joint matrix. During the training process, the technology of supervised learning can be used to achieve the rapid convergence of the training result. The present invention can effectively reduce the Matrix operation complexity, simple system hardware implementation, optimized system performance.

Description

technical field [0001] The invention relates to the field of communication technology, in particular to a deep learning-based MIMO hybrid beamforming method. Background technique [0002] Multiple input multiple output (Multiple input multiple output, MIMO) technology is a key technology in the communication field. It is widely used due to the rich spectrum resources in the millimeter wave frequency band. However, there is a serious path loss in the millimeter wave frequency band. Therefore, it is necessary to implement MIMO technology in the millimeter wave frequency band. A beamforming method is employed to compensate for path loss. Because the all-digital beamforming method is difficult and expensive to implement in massive MIMO system hardware, while the analog beamforming method only supports single-stream data transmission and has low system performance for MIMO systems, so digital beamforming and analog beamforming are adopted The way of combination has become a rese...

AI Technical Summary

Problems solved by technology

At present, MIMO communication systems based on traditional algorithms have become mature and have achieved good results in engineering practice, but most of them are based on complex matrix operations and the complexity of system hardware implementation is high.

Images