Research On Source Number Estimation In Array Direction Of Arrival

As a method of signal processing technology,source number estimation continuous development in recent decades.It has become an important branch of the signal processing field,and it has a large number of applications in the fields of radar,communications,medical etc.All kinds of super-resolution spectrum estimation algorithm prerequisite in signal processing is that the number of sources is precisely known,if the number of sources is unknown,or the sources number we obtained is not the real numbers,then the performance of these super-resolution algorithms will reduce and even seriously affected the normal completion of DOA Estimation.This paper focuses on the source number estimation under non-uniform noise,the main work is listed as follows:1)First,we introduce information theory criteria in the condition of the white noise,then under colored noise background,we introduce source number estimation method based on the algorithm of Gail circle criteria guidelines and characteristic subspace projection.Analyze the performance and shortcomings of these sources estimate methods.2)Introduces a non-uniform noise suppression method based on the transformation matrix.Since this method reduces the utilization of array,we propose a new method for receiving data delay information by suppression of non-uniform noise,which effectively inhibit the effects of non-uniform noise by delay processing the received data,and it does not reduce the utilization of the array.Using the eigenvalues of the covariance matrix obtained by delay processing,we describes the source number estimation method based on diagonal loading technique.Since this method reduces noise-signal ratio,this paper propose a new source number estimation method using a nonlinear mapping function,the eigenvalues of the covariance matrix map to the new space by mapping function so that the signal eigenvalues and the noise eigenvalues are well separated,and finally estimate the source number.The advantages of this new approach are: stable performance at low signal to noise ratio,and its performance is not affected by changes in the number of source signals.3)Through computer simulations and measured data test verify the performance of source number estimation methods described in this paper.Computer simulation and measured data show that when the number of snapshots conditions is met,the sources number estimation using delay correlation processing based on nonlinear mapping can effectivelycomplete source number estimation and has a excellent performance.