Array DOA Estimation Based On Compressed Sensing In Vehicle Integrated System

The rapid development and integration of communication technology and radar technology has enabled it to expand from military use to commercial use,and to today’s civilian vehicle radar communication integration.This system is cleverly applied to the Internet of Vehicles,and uses the integrated signals to synthesize estimated parameter information,such as direction of arrival angle,distance,and speed,and then locate the vehicle’s position.For this reason,the accurate estimation of the direction of arrival angle in the Internet of Vehicles is very important.The rapid movement of vehicles and the complex and changeable surrounding environment reduce the performance of traditional algorithms.In view of the many potential advantages of CS reconstruction algorithm,this paper proposes a solution to the problems of DOA estimation of spatial sparse partitioning and reconstruction algorithms under the uniform circular array and L-shaped array.The main points of the research are summarized as follows:(1)At present,the establishment of the sparse model simply adopts the equal angle or equal sinusoidal thinning method.However,in the normal and end-fire directions,the orthogonality of the two is different,which results in different estimation effects.To this end,under the uniform linear array model,a sparse fusion partitioning model is proposed.Combined with OMP algorithm for verification and comparison,the advantage of this model is not obvious when the SNR is relatively low,but it is significant when the SNR is relatively high.(2)Under the uniform circular array estimation,the array flow pattern matrix does not conform to the Vandermonde matrix structure,which makes it impossible to estimate directly.Thus,the introduction of the mode space conversion is considered to meet the structural requirements.What’more,in view of the scarce research on compressed sensing algorithms under the uniform circular arrays,this paper proposes a new UCA-OMP estimation method.This algorithm performs SVD processing on the transformed sparse model,adopts the OMP algorithm to sparsely solve the azimuth angle,and finally calculates the elevation angle estimation value based on the obtained azimuth angle estimation value to achieve accurate signal reconstruction.(3)Under the L-shaped array estimation,the two-dimensional dictionary matrix is directly constructed,which makes the computational complexity too high;when the amplitudes are close,the angle matching is easy to make mistakes,so that it cannot be directly applied to the actual in-vehicle system.As to this problem,this paper proposes a new method for 2D DOA estimation under the L-shaped array.This method uses the singular value decomposition method to reduce the dimension,and defines the spatial synthesis angle to construct a redundant dictionary,which drops the computational complexity twice.In addition,adopts the subspace projection method to achieve angle matching to reduce the pairing error rate and obtain the correct two-dimensional value.Simulation experiments show that using the array flow pattern matrix constructed by the sparse fusion partitioning method proposed in Chapter 3 as the basis of the algorithm,its estimated performance is better than the equal angle and equal sinusoidal method.The two-dimensional estimation method under the uniform circular array proposed in Chapter 4 does not require spectral peak search,the amount of calculation is small,and the angle estimation results of the 2D coherent signal and the non-coherent signal are also closer to the true value.The two-dimensional estimation method under the L-shaped array proposed in Chapter 5 can not only solve the huge calculation problem of constructing a two-dimensional dictionary matrix,but also avoid the two-dimensional spectral peak search and effectively guarantee the performance of angle matching.