Research On New Massive MIMO OFDM Channel Estimation Method

The technology of enlarging the antenna scale of base stations is called Massive Multiple-input Multiple-output(Massive MIMO),which can provide services for multiple single-antenna users at the same time.Orthogonal Frequency Division Multiplexing(OFDM)technology has high spectral efficiency and excellent anti-interference ability.In view of the increasingly precious spectrum and energy resources,it is an inevitable trend to combine MIMO and OFDM technologies,and the combination of the two technologies is the key of 5G technology.Accurate Channel State Infirmation(CSI)can give full play to the tremendous advantages of MIMO OFDM technology.Pilot-based channel estimation method is mainly used in actual transmission environment.The channel estimation method with less pilot cost is particularly important to improve the capacity and accuracy of Massive MIMO channel.In recent years,due to the precious pilot resources in Massive MIMO channels,the actual transmission channel has sparse structure characteristics.Compressed sensing theory is widely used in MIMO OFDM systems,which has significant effect in reducing pilot consumption.The existing sparse channel estimation algorithms based on Compressive Sensing(CS)not only have long running time,but also have high estimation complexity and high pilot consumption,which hinder the improvement of system performance.This thesis mainly focuses on two aspects of improving the pilot design and channel estimation methods of Massive MIMO OFDM system:(1)The channel estimation methods in this thesis are established: According to whether pilot symbols are used or not,the channel estimation methods are divided into three categories: pilot-aided estimation method,blind estimation method and semi-blind estimation method.The practicability of these three kinds of algorithms is discussed in detail.The commonly used pilot-based estimation method is established as the research direction of thesis paper.(2)The thesis discuss the theory of compressed sensing comprehensively,then extend MIMO to Massive MIMO technology,demonstrate Massive MIMO is the development trend of wireless communication system from system capacity,elaborate the basic idea of OFDM,and prove the advantages of combining the two.The traditional channel estimation methods of MIMO include Least Squares(LS)and Minimum Mean Squared Error(MMSE),which are not suitable for Massive MIMO system.Then,two greedy channel estimation methods are analyzed: Orthogonal Mathing Pursuit(OMP)and Sparsity Adaptive Matching Pursuit(SAMP).The shortcomings of OMP and SAMP estimation methods are discussed,which pave the way for the improvement methods proposed in the next chapter.(3)Based on the theory of structured compressed sensing,this thesis explores the structural sparseness of the channel and analyses the restoration method under the multiple measurement vector(MMV)model.In addition,an improved structured Othogonal Matching Pursuit(SOMP)method is proposed based on CS non-orthogonal pilot design.Compared with the traditional compressive sensing method in Chapter 2,simulation results show that the proposed method has a reliable performance improvement.(4)By analyzing the common sparsity of virtual angle domains between different subcarriers,a virtual Angle domain structured Othogonal Matching Pursuit(VA-SOMP)method is proposed.Moreover,MMV is extended to generalized multiple measurement vector(GMMV),and a nonorthogonal pilot scheme is designed.According to the simulation results,the VA-SOMP method based on this design reduces the time-slot consumption compared with the traditional method,and is almost close to the ideal curve.