Research On Two-dimensional Direction Of Arrival Estimation For Uniform Rectangular Array

As one of the key technologies for future wireless communication systems,massive multiple-input multiple-output(MIMO)technology,which can fully tap the spatially dimensional resource and improve system throughput,has attracted much attention of scholars both homeand abroad in recent years.The two-dimensional direction of arrival(2D DOA)estimation technology,one of the technologies of massive MIMO,holds very considerable research significance.At first,in massive MIMO system,the efficient 2D DOA estimation allows three-dimensional(3D)beamforming to accurately deliver the signal to target user by distinguishing users from horizontal and vertical dimensions.Second,2D DOA estimationcan provide base station with the accurate information of source,so that the optimal downlink precoding design will become more simple and feasible.Third,owing to different users with different 2D DOA,the new strategy can be provided to eliminate the pilot contamination.However,along with the dramatically increase of the number of base station antennas,the computational complexity of 2D DOA estimationwith uniform rectangular array will increase sharply.At the same time,under the conditions of low signal-to-noise ratio(SNR)and limited number of snapshots,some of 2D DOA estimation methods will exhibit poor performance or fail to estimate the 2D DOA for the coherent sources.Even if some 2D DOA estimation methods based on “dimensionality reduction” can work well but require additional pair-matching.Due to the reasons mentioned above,the thesis focuses on these problems and provides corresponding solutions.The content of the thesis can outline as follows:1.Based on the second order statistics of the received data under the uniform rectangular array,a local searching-based method is proposed for 2D DOA estimation.First,it takes advantage of the real eigenvector approximation to estimatesteering matrix to reduce the number of times of total least square(TLS),the initial 2D DOA estimation of incoherent sources will be achieved with lower computation.Second,instead of angle traversal searching,the 1D local searching is exploited to optimize initial value,which not only efficiently improve the precision of angel estimation in the range of low signal to noise ratio but also avoid the drawback of heavy computational complexity due to traversal searching.2.A sparse reconstruction 2D DOA decoupling estimation method is proposed for uniform rectangular array,which successfully avoids the extra angle matching problem.Firstly,the vectorization operationand the matrix kronecker product in the method is effectively exploited to decouple the angle parameters,and then the 1D over completed dictionary is constructed to sparse the signal model.Finally,the constraint relationship of two angle parameters is deduced to realize the 2D DOA automatic pairing estimation.The proposed method only requires one timeof sparse solution,and the redundancy of required dictionary is approximately equivalent to 1D over completed dictionary.The method effectively avoids angle matching,and hence the computational complexity is effectively reduced.The proposed method exhibits better robustness and angle resolution under the circumstance of the coherent and incoherent sources.