Research On Improved DOA Algorithm Based On Large-Dimensional Scenarios

In recent years,large-dimensional arrays such as Multiple-Input Multiple-Output(MIMO)arrays and Ultra-Dense Network(UDN)have developed rapidly.In largedimensional scenarios,the number of antennas/sensors is in the same order of magnitude as the number of observation samples.But the condition of traditional array signal estimation is that the number of array elements should be much smaller than the number of samples.It is of great practical significance to study the problem of array signal processing in large-dimensional scenarios.Direction of Arrival(DOA)estimation is an important branch in the field of array signal processing.Subspace-based DOA estimation algorithms have been widely used as their super-resolution property,and its performance depends on the time-averaged sample covariance matrix.In large-dimensional scenarios,when the sample covariance matrix of the received signal cannot be used as an estimator of the statistical average covariance matrix,and the performance of the DOA algorithm would be degrade dramatically.In the field of wireless communications,the sample correlation matrix of the channel received data also affects the performance of channel estimation.For the above problems,largedimensional random matrix theory and G-estimation theory provide a theoretical basis and estimation framework for parameter estimation in large-dimensional scenarios.Based on large-dimensional random matrix theory and G-estimation theory,we study DOA estimation and channel estimation in large-dimensional scenarios.The main work is as follows:(1)Aiming at the problem of the DOA estimation in large-dimensional scenarios,the empirical spectral distribution of the sample covariance matrix is analyzed based on the Stieltjes transformation of large-dimensional random matrices,the empirical spectral functions of eigenvalues and eigenvectors are constructed,and the empirical spectrum in large-dimensional scenarios is determined.The infimum of the ratio of the matrix dimension and the number of samples that can be suitably applied to DOA estimation under large-dimensional scenarios is analyzed,and a subspace-based DOA estimator that satisfies the infimum condition and converges to the non-random quantity in G-estimation theory is proposed.The simulation results show that the subspace-based DOA algorithm based on large-dimensional random matrix theory and G-estimation theory has better resolution and estimation performance under small samples scenario.(2)Aiming at the improvement on the key ideas of the large-dimensional subspacebased DOA estimation algorithm,the correction coefficients in the cost function are extracted for analysis,and the modification method of the sample covariance matrix by the correction coefficients is also verified simulated.The simulation results show that,compared with the traditional subspace-based DOA algorithm,the large-dimensional DOA estimation algorithm weights the eigenvectors of the sample covariance matrix,so that the DOA estimation maintains a better estimation performance with a small number of samples.(3)Aiming at the problem of channel estimation in large-dimensional scenarios,a channel estimation algorithm based on corrected sample correlation matrix is proposed by analyzing the performance of channel estimation in large-dimensional scenarios.The proposed algorithms with the correction coefficients of G-estimation is able to improve the estimation performance.The simulation results show that the algorithm can effectively improve the channel estimation performance in the case of small samples and low signal-to-noise ratio.