Low Complexity Channel Estimation In 3D MIMO Systems

As one of the core technologies of 5G,3D MIMO(Three-Dimensional Multiple-Input and Multiple-Output)can significantly improve the system capacity and reliability.Channel estimation problem in 3D MIMO systems can provide channel parameters for diversity combining,coherent detection and decoding.The estimation performance directly affects the performance of the whole receiver system.In 3D MIMO systems,with the increase number of antennas and the reduction of the spatial distance,there is a certain correlation between antennas.The Weiner filtering algorithm was famous for its superior estimation performance.However,its computational complexity is very high due to the large number of antennas in 3D MIMO systems.How to take fully advantage of 3D MIMO characteristics to decrease the complexity of the algorithm is worthy of further study.In order to reduce the pilot overhead and make fully use of the sparsity of the 3D channel,the compressive sensing theory was applied to the process of 3D MIMO channel estimation.The design quality of the measurement matrix plays an important role on the estimation performance.To get the optimal allocation scheme,exhaustive search methods are usually required.However,the computational complexity becomes unsustainable when the number of antennas increases.A design method of measurement matrix with low complexity is urgently needed.In this thesis,we studied the Weiner filtering algorithm and compressive sensing based channel estimation in 3D MIMO systems,respectively.The main work of the research is to reduce the computational complexity of the two kinds of channel estimation methods.The main contributions of this thesis are summarized as follows:1)By using the spatial correlation between the transmit antennas,this thesis inotrduced a cascaded MMSE(Minimum Mean Square Error)estimator in spatial space for 3D MIMO system.The cascaded MMSE estimator decreases the complexity of MMSE channel estimation by replacing the correlation matrix with the correlation matrices in azimuth and elevation directions,respectively.The simulation results show that the MSE(Mean Square Error)and BER(Bit Error Rate)performances are significantly improved over LS(Least Square)algorithm and RBFN(Radial Basis Function Network)algorithm,and had comparable MSE and BER with MMSE channel estimation.Meanwhile,the proposed cascaded MMSE channel estimation method significantly decreases the computational complexity compared with the two-dimensional MMSE estimator.The simulation results also demonstrate that the channel estimation in spatial domain can be decomposed into horizontal domain and vertical domain respectively.2)This thesis makes fully use of the advantages of quantum Grover algorithm in large data search,and applies the quantum search algorithm to the measurement matrix design.The most commonly used method of the pilot allocation scheme is uniform distribution,which has a poor performance.However,the optimal allocation usually requires ergodic search whose computational complexity becomes unbearable when the number of antennas increases.This quantum Grover algorithm based scheme this thesis introduced has much lower computational complexity with respect to the traversal search method.Compared with the uniform distribution,the estimation performance is significantly improved.