Research On DOA Estimation Algorithm Based On Deep Learning And Received Signal Strength

Direction of arrival(DOA)estimation is a key research direction in the field of array signal processing.In the past two decades,the research on DOA estimation algorithm has been the focus of attention from all walks of life,and its theoretical research results have also been widely used in sonar,radar,radiation source positioning,military communications and many other fields.Taking an overall view of the currently published DOA estimation algorithms,part of which is based on physical information such as phase or time,and part of which is based on physical information such as power or amplitude.The latter can be collectively referred to as the method based on received signal strength(RSS).This kind of method is expected to become the main driving force for the development of future DOA algorithms because of its advantages of low complexity such as not relying on tight time synchronization and higher sampling rate.Although there are few related researches at present,and the existing algorithms still have some challenges,with the emergence of new technologies and new theories,the DOA estimation algorithm based on RSS will show more possibilities and solutions.Based on the analysis and summarization of the latest achievements in related fields at home and abroad,this paper combines the thought of deep learning(DL)and the theory of compressed sensing(CS)to study the RSS-based DOA estimation algorithm from the point of view of antenna radiation pattern,which refers to the radiation pattern of signal source.The main work of this article is as follows:Firstly,this paper introduces the theoretical basis of compressed sensing,deep learning and convolutional neural network(CNN),and makes a purposeful analysis for the main research contents in these theories.Combined with practical application cases,the applicable environment,advantages and disadvantages of these theories are discussed,which makes matting for the subsequent research work.Secondly,this paper conducts a study on the measurement process of the antenna radiation pattern.Aiming at the phenomenon that the existing measurement methods have high complexity and low efficiency.This paper introduces the theory of compressed sensing and proposes an antenna radiation pattern reconstruction algorithm based on CS to reduce the number of sampling points needed in the measurement process.In addition,in order to overcome the difficulty which random measurement matrix is not easy to be realized by hardware in practical application.This paper also designs a construction method of deterministic measurement matrix based on m-sequence.The simulation results show that the proposed algorithm can reconstruct a complete antenna radiation pattern by using partial sample values,the number of which accounts for 50% of the total data.And the results are highly close to the reconstruction effect of random measurement matrix.Finally,aiming at the problem that the current RSS-based DOA estimation methods have low accuracy and small application range.This paper designs a DOA estimation method based on CNN and RSS.By analyzing the relationship between the radiation pattern of the signal source and the RSS of the sensors,a DOA classification model based on CNN is established,which realizes the high-precision DOA prediction in the long-distance or outdoor environment,and provides more reference for the research of DOA estimation based on RSS.