Research On Trainning Sequence Optimization Algorithm In Massive MIMO System

5G is a new generation of mobile communication system,which is based on the requirement of mobile communication after 2020.The core technologies include wireless transmission technology with high efficiency and high density wireless network technology.The key point is the massive MIMO Technology.Massive MIMO system has the advantages of high spectral efficiency and energy efficiency.Since it was proposed,it has been widely and deeply studied and it is the focus of research in the field of communication.In this paper,we study the channel estimation technology of massive MIMO system,and study the optimization of downlink training algorithm in FDD MIMO system.It is necessary to study the performance limitations of traditional training algorithms in large-scale MIMO systems.For traditional training algorithms,there are performance limitations in large-scale MIMO systems.This paper studies the open-loop training algorithm for the applicability of large-scale MIMO system,the minimum mean square error estimation on training sequence structure of the ring channel single shot under conditions is deduced,according to the training sequence after optimization of the ceiling effect through the theoretical and practical simulation proved the existence of open-loop single shot training method the training sequence is fixed,the system in the large scale MIMO system normalized SNR will be limited.The use of open-loop training methods can alleviate the memory performance limitations of existing open-loop single shot training,using Calman filter to predict the large scale MIMO system can improve the channel estimation accuracy of channel information.In order to solve the problem of training sequence calculation in the base station of large-scale MIMO systems,this paper focuses on the closed loop memory training algorithm in large-scale MIMO systems.Study the channel estimation performance of the best training sequence structure design method,the MSE minimum complete feedback signal through the training signal for training set and high performance,improve the channel estimation performance.Simulation results show that the closed loop method of memory training can be estimated more accurately in the channel only add a few bits of feedback overhead,when all the training signal structure user feedback,the system further downlink can get higher channel estimation performance.In this paper,a closed loop memory training algorithm with power allocation strategy is proposed innovatively.The base station can achieve more accurate downlink channelestimation without the long term statistical information of the channel.The mean square error performance of the new algorithm is evaluated by computer simulation.The proposed algorithm can achieve better tradeoff between system performance and link overhead,and is suitable for the training design problem in large-scale MIMO systems.