Deep learning MIMO system signal detection method and system

Abstract

The invention discloses a deep learning MIMO system signal detection method and system, and the method comprises the steps: S1, carrying out the preprocessing of a received signal, including high-order modulation signal decoupling and signal sorting, obtaining a signal with a time sequence relation after preprocessing, and inputting the signal into a neural network decoder for training; s2, pruning the trained neural network decoder by using a channel pruning method; and S3, the neural network decoder carries out multi-category prediction on the preprocessed signal, and enables the predicted category to correspond to a modulation signal constellation diagram to realize prediction of the sending signal. According to the method, low-order decoupling and signal reordering processing are carried out on input high-order signals, so that the false detection problem of multi-classification detection tasks and the prediction order problem of tasks with historical time prediction result dependence are reduced.

Description

technical field [0001] The invention belongs to the technical field of wireless communication, and in particular relates to a deep learning MIMO system signal detection method and system. Background technique [0002] In the communication system, signal detection, as a key part, plays a very important role in the communication quality of the communication system. [0003] Traditional signal detection can be divided into two categories: linear detection and nonlinear detection. Linear detection, such as zero-forcing detection, maximum mean square error detection, has low computational complexity, but low detection accuracy; nonlinear detection, such as maximum likelihood estimation, spherical decoding, has high detection accuracy, but computational complexity also rose. It can be seen that in traditional signal detection, it is difficult to achieve both lower computational complexity and higher detection accuracy. How to solve this contradiction has become a research focus ...

AI Technical Summary

Problems solved by technology

Linear detection, such as zero-forcing detection, maximum mean square error detection, low computational complexity, but low detection accuracy; nonlinear detection, such as maximum likelihood estimation, spherical decoding, high detection accuracy, but computational complexity also rises

It can be seen that in traditional signal detection, it is difficult to have both low computational complexity and high detection accuracy. How to solve this contradiction has become a research hotspot in current signal detection.

Images