| In 5G(Fifth Generation)wireless communication system,users urgently need high speed,low latency and high reliability wireless data transmission services,but often face the problem of frequency resource shortage,so expanding frequency to higher frequency bands has become one of the main directions for 5G communication.Millimeter Wave(mmWave),with its huge bandwidth sufficient to provide gigabits per second data throughput,is one of the core technologies for next-generation cellular wireless communication technologies.At the same time,by combining millimeter wave communication with Multiple Input Multiple Output(MIMO)system,not only can the spectrum utilization of millimeter wave communication be effectively improved,but also can overcome the severe path loss caused by high frequency communication.In order to ensure the efficient combination of millimeter wave communication and MIMO system,both the base station and the user side need good channel state information,and in the 5G millimeter wave high-speed mobile scenario,the high mobility of the user side leads to more complex channel state,and the effect of Doppler shift brings more challenges for channel estimation.Therefore,this thesis will focus on the time-varying channel estimation problem in millimeter wave MIMO systems,and investigate the channel estimation algorithms in different communication scenarios by combining the mathematical theory of tensor decomposition with the requirements of low guide frequency overhead,low complexity,and high robustness in time-varying channel estimation algorithms.The main contents of this thesis include:(1)For Frequency Division Duplex(FDD)system,the downlink time-varying channel estimation is mainly studied,and the user side can feed the channel state information to the transmitter side.In the point-to-point MIMO system,the user side uses a two-stage channel estimation method to divide the channel estimation into two parts:angle estimation,Doppler Shifts and complex gain estimation,and uses tensor decomposition to assist the channel estimation.In the first stage,the received signal is modeled as a third-order tensor,and a structured tensor decomposition method is proposed to estimate the angle information for the problem of low stability of the traditional tensor decomposition method.This method uses the Vandermond structure of the tensor factor matrix and uses a pure algebraic method to recover the parameters,avoiding the instability of the traditional tensor decomposition method that requires iterations,and achieving the effect of improving stability and enhancing performance;in the second stage,the Doppler Shifts and the complex gain are recovered from the derived angles and the received signals.(2)Extending to the multi-user MIMO scenario,in the Time Division Duplex(TDD)system,the base station side can convert the uplink channel into the downlink channel using channel reciprocity,so the uplink channel estimation is mainly studied.The two-stage channel estimation method is used at the base station to divide the channel estimation into two parts:angle estimation,Doppler Shifts and complex gain estimation.In the first stage,a dictionary-based tensor decomposition method is proposed to estimate the angle in response to the high complexity of the traditional tensor decomposition method,which takes advantage of the sparsity of the millimeter wave channel in the angle domain and greatly reduces the number of iterations in the tensor decomposition calculation process,thus achieving the effect of reducing the complexity;In the second stage,the Doppler Shifts and the complex gain are recovered from the obtained angle and the received signal.In this thesis,we use tensor decomposition to perform channel estimation,which can effectively reduce the pilot overhead of channel estimation,and further improve the performance of the traditional tensor decomposition method by using the sparsity of the tensor factor matrix and the Vandermond structure. |
PDF Full Text Request