Research On Advanced Receiver For Large Scale Multi-antenna System

With the emergence of new mobile application scenarios and demands,the 5th Generation Mobile Communication Systems(5G)has become a necessary solution for a large number of wireless device access and massive mobile data transmission in the next decade.As one of the core technologies of 5G,large-scale multi-antenna system can greatly improve the throughput and transmission rate of communication systems,and can be combined with other core technologies of 5G such as millimeter wave and Orthogonal Frequency Division Multiplexing(OFDM)to meet a large number of high-speed data transmission needs.In the communication system,the performance of the receiver directly determines the performance of signal recovery.However,as the number of antennas in the large-scale multi-antenna system increases,the complexity of the receiver increases,and the cost of the RF link increases.Whether the algorithm can be applied in reality is yet to be verified.This paper is aimed at large-scale multi-antenna systems,using a combination of simulation and over-the-air measurement to study advanced receivers.First of all,this paper studies and implements a large-scale multi-antenna receiver based on traditional methods.After investigating the traditional receiver algorithm,the low complexity LMMSE channel estimation algorithm 3W-LMMSE and 12W-LMMSE are designed.Based on the multi-core general server architecture 5G prototype verification platform RaPro,introduction of RaPro,top-level design of receiver and specific implementation of the designed algorithm are presented.The simulation results show that the proposed 3W-LMMSE and 12W-LMMSE channel estimation algorithms have obvious performance improvement compared with the traditional LS channel estimation algorithm,and are robust to channel path number and SNR parameters.The over-the-air measurement results verify the results,as well as the accuracy and real-time nature of the two methods.Secondly,this paper studies the large-scale multi-antenna advanced receiver based on Expectation Propagation(EP)algorithm.After reviewing the system model and augmented system model of the large-scale multi-antenna single-user millimeter-wave communication system,the basic principles of the EP algorithm are introduced,and then the a posteriori mean and covariance of the augmented millimeter-wave channel is derived.The iterative process of the EP algorithm is derived according to the moment matching condition and the EP-based channel estimation algorithm is given.The simulation results show that the EP-based channel estimation algorithm has obvious advantages compared with other existing algorithms and can adapt to lower compression ratio.Finally,this paper studies the noval deep learning based receivers.After introducing depth learning and model-driven deep learning,the proposed ComNet OFDM receiver is introduced in detail from the overall architecture,channel estimation subnet and signal detection subnet,and extended to multiple input multiple output OFDM system.Simulation experiments show that the ComNet OFDM receiver has a certain performance improvement compared with the traditional receiver,and it is robust to changing the training channel,reducing the pilot placement,and the signal-to-noise ratio mismatch.The over-the-air test also achieved better performance than the traditional method in three scenarios.