Research And Design Of DOA Estimation System

Direction of Arrival(DOA)estimation has always been the focus of research in the field of array signal processing,and has potential applications in the fields of communications,exploration,biomedicine,and radar.Although in the current stage,DOA estimation has achieved a lot of research results,there is still a trade-off between estimation accuracy and computational complexity in DOA estimation.In this dissertation,we will use the convolutional neural network with super non-linear fitting ability to propose a fast DOA estimation method with high estimation accuracy,and build a DOA estimation experimental system based on software radio,and verify the estimation performance of the proposed algorithm.The main work of this dissertation is as follows:First,the basic principles of DOA estimation,software radio and convolutional neural networks are introduced.In view of the large amount of calculation and low estimation accuracy of traditional DOA estimation algorithms,a DOA estimation algorithm based on Adam-Dropout-CNN is proposed.This algorithm uses a Dropout layer to avoid overfitting during convolutional neural network training.And through the Adam optimizer to optimize the convolutional neural network to improve the accuracy of DOA estimation.The simulation results show that,compared with the MUSIC algorithm and the ESPRIT algorithm,this algorithm can achieve high-precision and fast DOA estimation under the condition of low signal-to-noise ratio and few elements.Then,based on the theoretical basis of software radio technology,the application potential of software radio technology in wireless direction finding is analyzed.At the same time,software radio hardware platform NI USRP-2920 and software radio development software GNU Radio and Lab VIEW platform are used to design a software-based radio technology.Radio’s DOA estimation experimental system.Finally,the MUSIC traditional algorithm is used to verify the estimation performance of the DOA estimation experimental system,and the estimation error of the experimental platform is analyzed and corrected.At the same time,the AdamDropout-CNN algorithm proposed in this dissertation is also carried out in the DOA estimation experimental system.verification.The experimental results show that under the conditions of low signal-to-noise ratio and small number of elements,the proposed algorithm can achieve high-precision and fast DOA estimation within the error tolerance range.In the case of changing the azimuth angle of the signal source many times,the DOA experimental system built in this dissertation can distinguish the azimuth angle of the signal source.The DOA estimation system designed in this dissertation has been studied from two aspects: algorithm theory simulation and experimental verification.It proves the feasibility and practicability of the DOA estimation system designed in this dissertation,and it has certain potential value in practical applications.