Research And Implementation Of MIMO-OFDM Advanced Receiver For Intelligent Communication

Currently,5th Generation Mobile Communication Systems(5G)has entered the commercial stage.Multiple Input Multiple Output(MIMO)technology can improve the throughput and spectrum efficiency of wireless communication systems,and therefore has become the one of the key technologies.With the emergence of new communication scenarios and the massive access of mobile terminal equipment,data traffic has exploded.Traditional communication systems have encountered bottlenecks in which the new channel environment is too complicated to model and massive data signals are difficult to process in real time.In recent years,deep learning(DL)has received widespread attention because of its powerful function simulation and the ability of high-speed parallel processing of big data.Therefore,the research on the application of DL in the physical layer of wireless communications is rapidly emerging.Intelligent communication combined with Orthogonal Frequency Division Multiplexing(OFDM)technology and MIMO technology has become hot research direction of 5G and even the future mobile communication development.This article mainly studies the MIMO advanced receiver based on DL.It uses simulation and air interface test to conduct further analysis and research on the performance of the receiver..First,this article studies the key technologies in MIMO receiver.After investigating the key algorithms in traditional MIMO systems,including traditional channel estimation and signal detection algorithms,three basic neural network structures are introduced: fully connected networks,convolutional neural networks and recurrent neural networks,and neural network optimization methods also mentioned.Then,the modular application of DL technology in the physical layer of wireless communication is studied,including frame synchronization,modulation pattern recognition,channel coding,and signal detection.It also further explores a new architecture that replaces the complete SISO-OFDM receiver with neural networks,and the possibility to extend to MIMO systems.Next,this paper studies the design and implementation of a MIMO-OFDM OMNet advanced receiver based on a fully connected neural network architecture.After introducing the basic principles and advantages of OFDM modulation technology,the article describes overall design idea which combines DL technology and communication knowledge,and detailed description of the specific structure and working principle of its channel estimation module and signal detection module,the subsequent simulation results reveal the superiority of OMNet receiver compared to traditional receivers.Then this article deploys the OMNet receiver on Ra Rro’s rapid development prototype verification platform,and gives a detailed introduction to the top-level architecture design and software implementation of the receiver.The air interface test results further prove that the OMNet receiver is superior to traditional receivers.Finally,the advantages and limitations of OMNet receivers’ extension to massive MIMO systems are analyzed from three aspects: network architecture,computing speed and training speed.Finally,this paper studies the channel estimation algorithm based on DL image processing and the signal detection algorithm driven by DL model.First introduced the application results of convolutional neural network in image processing,and analyzed the basic principles and feasibility of transferring the above application results to the channel estimation module of MIMO system.Then based on this,the SR-Net channel estimation module of OFDM-MIMO system was designed,then analyzes the non-applicability of the AMP algorithm in a non-linear,ill-conditioned channel matrix.Based on this,the OAMP algorithm is proposed and the principle of the algorithm is explained in detail.The OAMP-Net signal detector combined with DL technology innovatively expands the iterative process of the algorithm into a neural network,and further optimizes the signal detection performance by setting fewer training parameters.Finally,the performance of SR-Net channel estimation module and OMP-Net signal detection is obtained through system simulation,which illustrates that the above algorithms have obvious advantages compared with the performance of traditional communication algorithms.At the same time,it is concluded that the above algorithm has fewer training parameters,simple network structure,and can be applied to massive MIMO systems.