| With the advent of the 5th generation mobile communication technology(5G)era and the gradual rise of artificial intelligence technology,the combination of artificial intelligence and communication technology has become a trend.The 5g extended Pico site connects multiple RRUs(Remote Radio Units)to a BBU(Base Band Unit).The uplink signals received by multiple RRUs are combined into a base station signal in some way and sent to the BBU for processing.In this thesis,the deep learning method is applied to the preprocessing of RRU uplink signal.The purpose is to filter the part of a RRU received signal that does not contain useful signal and improve the signal-to-noise ratio of the combined uplink signal.In this thesis,multiscale convolution is applied to one-dimensional convolution neural network,and a 1D-MCL-MS(One Dimensional Multiscale Convolution Multiple Signals)classification model is proposed.Compared with the traditional CNN(Convolutional Neural Network)model,1D-MCL-MS model has higher classification accuracy;Compared with the traditional Fourier noise processing method,it has lower time delay.Based on 1D-MCL-MS model,this thesis designs and implements RRU uplink signal preprocessing system for extended Pico site.The RRU data preprocessing system based on 1D-MCL-MS model implemented in this thesis includes RRU end band signal identification unit,filtering unit,uplink signal merging unit and other modules.The band signal identification unit adopts the 1D-MCL-MS classification model proposed in this thesis,which can identify and classify the band signal in the first several sampling points in the time domain,and distinguish the band signal and band noise.According to the classification results,IIR(Infinite Impulse Response)digital band stop filter is used for noise reduction.The merging unit adopts the selective merging method to integrate the signals of multiple RRU channels into the BBU.Selective merging has the characteristics of low delay and low complexity,which can further reduce the overall delay of the system.This thesis discusses the design and implementation of the system in detail,and completes the function test and performance analysis.Overall,the system designed and implemented in this thesis has the characteristics of high reliability and low delay,and the overall performance has been improved. |
