Research On Estimation Technology Of Communication Signal's Coming Wave Direction Based On Deep Learning

Direction of arrival estimation(DOA)plays a very important role in electronic reconnaissance countermeasures,radio spectrum monitoring,mobile communications,radar,sonar and other fields,and is an important branch in the direction of array signal processing.In the current communication environment,there is a demand for a large array.In a large array system,it will bring huge challenges to the traditional direction of arrival estimation algorithm,and face problems such as large amount of calculation and high-precision trade-offs.Traditional arrival wave direction estimation algorithms have poor adaptability in the actual environment,because this type of algorithm is an algorithm based on mathematical expressions and needs to make many assumptions about the environment.When the actual environment is difficult to reach the target conditions,the estimated performance cannot meet the actual needs.,And the anti-noise ability is poor.Deep learning can realize the analysis and processing of large-scale data.It is a data-driven algorithm without making assumptions about the environment.Estimating the direction of arrival of communication signals based on deep learning will greatly improve the calculation efficiency of estimation of the direction of arrival.Through data training of multiple signal-to-noise ratios,the anti-noise ability of the system can also be improved.In this case,how can we use the advantages of deep learning to solve the problems of slow calculation efficiency of traditional arrival direction estimation algorithms and unsatisfactory performance of low signal-to-noise ratio? Based on the latest research,this article has done the following work:(1)Research the mathematical models of array signals and coherent signal sources,and analyze the basic components of neural networks.(2)Research some traditional arrival wave direction estimation algorithms,and propose some improved algorithms based on the shortcomings of traditional algorithms.(3)Based on deep learning to estimate the direction of arrival of uniform linear arrays,the CNN-Line DOA estimation algorithm is proposed,and compared with the traditional MUSIC algorithm in many aspects,the feasibility of the algorithm is discussed.(4)The uniform linear array can only get the azimuth angle,but not the pitch angle of the communication signal.This article proposes a CNN-Circle DOA estimation algorithm based on a CNN network for a uniform circular array,which simultaneously estimates the azimuth and elevation angle of the communication signal through deep learning.This article will start from the research of historical DOA algorithm and use the advantages of deep learning to process big data.Based on two-dimensional space and three-dimensional space respectively,two methods for estimating the direction of communication signals based on deep learning are proposed to improve the computational efficiency and anti-noise ability of the algorithm.,It provides a feasible method for the deep learning algorithm to add the direction of arrival estimation.