| Massive MIMO systems employ hundreds of antennas on the base station,which will dramatically improve the diversity gain and the multiplexing gain of system,thus greatly improve the capacity of system.Suppose the number of users is fixed,when the antennas of BS grow toward infinity,channel of different users will present orthogonality,and the noise of system will also disappeare.Hence,simplest linear precoding in massive MIMO system can realize the capacity of shannon theorem.While in the real communication system,taking the hardware complexity and the energy consumption into consideration,the number of antenna cannot grow without limitation.When the antenna reached saturation,additional power consumption cannot improve system capacity.Therefore,how to select the antenna effectively and use them to communication efficiency is one of the research problem of massive MIMO system.In this article,we focus on the antenna selection related issues of massive MIMO system and carried out some related research,the main research contents are as follows:(1)Massive MIMO antenna selection technology.First,we researched the traditional antenna selection algorithm including optiaml antenna selection,normal based antenna selection,correlation based antenna selection,increase/decrease antenna slection.Based on this,we proposed sliding Window base fix-length antenna selection and non-fix-length antenna selection algorithm.Virtually the fix-length antenna selection algorithm is based on the principle of decrease progressively,remove the first or last antenna from the channel matrix each step,which is low complexity.The non-fix-length antenna selection algorithm will determine the window length based on the system energy efficiency,and goto fix-length antenna selection then.The algorithm we proposed has low complexity and it's superior to random selection algorithm.(2)Massive MIMO linear precoding technology.First,we analysis the system performace of MF,ZF,MMSE,BD precoding.As for the MMSE precoding has to calculate the pseudo inverse of the matrix which is hardware inefficient,we proposed a matrix inverse approximation algorithm that utilizes the Chebyshev iterative algorithm to optimized the initial value of Neumann series.As the simulation result indicated that approximation algorithm can reach a better MSE and system performance with less number of items.Then,based on the sliding window scene,we came up with a precoding recursive algorithm with removing antenna form the sliding window.The recursive algorithm can work perfectly with lower complexity meanwhile has the advantage of high hardware efficiency.(3)Massive MIMO channel estimation and pilot sequence optimization.We studyed the theory of LS and MMSE channel estimation and make some simulation of these two algrithem.By means of power allocation and pilot length optimize,we proposed the optimal pilot sequence for the 3D massive MIMO channel which can reduce the MSE of channel estimation.Simulation result show that the pilot sequence can reach a more accurate channel evaluation and can improved the SNR of receive signal. |
