Research On Direction Of Arrival Estimation Methods Based On Sparse Reconstruction

In the military field,detecting the position of the source has always been an important part of radar work.With the development of science,technology and hardware level,the spatial spectrum estimation based on the sensor array in the passive reconnaissance system has become the most commonly used Direction of Arrival(DOA)estimation method.With the deepening of research and the improvement of engineering,a large number of DOA estimation algorithms have been proposed and applied,and they are widely used in the fields of communication and radar.However,with the development of information technology and the increasing complexity of the electronic warfare environment,the requirements for high-resolution and stable multi-target direction estimation are getting higher and higher.Situations like low signal-to-noise ratio,low number of snapshots,coherent source interference and unknown model order have brought severe challenges to DOA estimation.DOA estimation methods supported by new theories are urgently needed to break through the limitations of traditional methods.Sparse reconstruction and compressed sensing theory has brought new development directions for direction finding.Due to its significant advantages in the fields of low snapshots,high resolution and decoherence,sparse methods have received extensive attention in recent years.While utilizing sparsity,improving the resolution and distinguishing coherent signals with adjacent spatial angles in the case of small snapshots and low signal-to-noise ratio is the most important topic in DOA estimation based on sparse reconstruction,which is one of the research purposes of this paper.The sensitivity of sparse reconstruction algorithms to noise is usually adjustable,which means that sparse reconstruction algorithms can be well combined with self-calibration methods to achieve robust DOA estimation.The research on the robustness of the algorithm is also a very important direction in the application of DOA estimation,and is another important research purpose of this paper.The main research work of this paper includes:First,this paper proposes a single snapshot DOA estimation algorithm based onl_p norm optimization.Starting from the selection of the sparse norm,with the goal of improving the resolution,thel_p norm is selected as the sparse metric.And the quasi-Newton method is used for non-convex optimization.The SWM formula is introduced and simplification is derived to reduce the time complexity of the algorithm.To achieve higher resolution of distinguishing closely spaced sources and accelerate algorithm convergence,an adaptivel_p norm is proposed as a selection method.The simulation results show that the proposed algorithm has good DOA estimation performance,and has higher resolution than other classic algorithms.Secondly,this article introduces three ways to use the multi-snapshot data that is suitable for the algorithm mentioned above without changing the core of the algorithm.Then,section IV extends thel_p norm optimization DOA estimation to multi-snapshot by using joint sparse.Moreover,a regularization parameter dynamic adjustment method is proposed,avoids the subjective selection of regularization parameters that most sparse reconstruction algorithms need,which affects the performance of the algorithm.In the severe DOA estimation conditions like the lower number of sensors and signal-to-noise ratio,a weighting algorithm is proposed to improve the performance.The simulation results show that the proposed algorithm has good DOA estimation performance,and has higher resolution than other classic algorithms.Finally,the sparse reconstruction algorithm is combined with the calibration algorithm to study the method to improve the robustness in practical DOA estimation.A robust DOA estimation algorithm is proposed for two common error or disturbance situations.One is to solve the problem of sensors position error in the actual DOA estimation scene,introduce genetic algorithm for array element position calibration,and select appropriate genetic algorithm parameters for the special application background of sensors position calibration.The simulation results show that the proposed algorithm can regulate the position error of the sensors to a certain extent in the case of known and unknown sources.For the more general data model under the condition of steering vector disturbance,a robust DOA estimation algorithm of alternating direction descent in combination with the DOA estimation algorithm mentioned above and the block coordinate descent is proposed.The simulation results show that the proposed algorithm can complete DOA estimation well even when the steering vector is disturbed.