Research On DOA Estimation Algorithm For Circular And Non-circular Mixed Signals

In the field of signal processing,array signal processing is a major research direction.Array processing is widely used in daily communication,sonar detection,seismic surveys and military radars.Direction of arrival(DOA)estimation is one of hot issues of array signal processing,which uses the special structure array to receive the signals and estimate the incident angle of the signals with designed algorithms.For DOA estimation,the model of the received data is usually related to the characteristic of the signals,which can be divided to circular signals and non-circular(NC)signals.The pseudocovariance matrix of non-circular signal is not zero,which can significantly improve the performance of DOA estimation.When the signal source emits non-circular signals,by utilizing the non-circularity of the signals,the received data can be augmented and the array aperture can be enlarged,which makes DOA estimation more accurate.In the actual communication system,common circular signals include QPSK signals,QAM signals,etc.Non-circular signals include BPSK signals,AM signals,MASK signals,etc.This paper concentrates on the researches of the circular and non-circular features,and proposes a series of DOA estimation algorithms for circular and non-circular signals,which owns rich research significance and practical application value.The main work of this paper is as follows:(1)The DOA estimation algorithms based on rotational invariance for circular and non-circular mixed signals with uniform linear array are studied,including Mixed signals Estimation Signal Parameters via Rotational Invariance Techniques(Mix-ESPRIT)algorithm and Mixed signal Propagator Method(Mix-PM)algorithm.Both algorithms reconstruct the received data by stacking the received signal matrix and its conjugate part.Among them,the Mix-ESPRIT algorithm uses the transformation relationship between the steering matrix of the reconstructed signal and the signal subspace to obtain the relationship between the matrix after block reorganization and the subspace recombination of the signal subspace by means of rotation invariance,and finally obtains the DOA estimates.The Mix-PM algorithm uses the relationship between the steering matrix of the reconstructed signal and the propagation operator to reorganize the matrix composed of the steering matrix and the propagation operator to obtain the DOA estimates.The proposed algorithms have good angle estimation performance,and can distinguish the circular and non-circular signals.Meanwhile,when the number of incident signals is more than the number of receiving array elements,the DOA estimation can still be performed on the incident signal via the proposed algorithm.In addition,as the number of non-circular signals in the mixed signal increases,the proposed algorithms' angle estimation performance is improved.(2)The DOA estimation algorithms based on Euler transform for circular and non-circular mixed signals with uniform linear array are studied,including Mix-Euler-ESPRIT algorithm and Mix-EulerPM algorithm.The proposed algorithm uses the Euler transform method to convert the complex received signal into a real number signal,and finds the real and imaginary parts of the received signal separately,and then uses the real part and the imaginary part to realize the extension of the received signal.The steering matrix is transformed via the properties of the steering matrix of the extended signal to obtain a matrix containing the information of the incident angle.In order to obtain the DOA estimation expression,the Mix-Euler-ESPRIT algorithm utilizes the transformation relationship between the signal subspace and the steering matrix,and then uses ESPRIT method to obtain the incident angle estimation,while the Mix-Euler-PM algorithm obtains the expression of the incident angle estimate by propagating the relationship between the propagation operator matrix and the steering matrix.The proposed algorithms obtain the DOA estimation angles of the circular signals in pairs,which is the basis for distinguishing between circular and non-circular signals.The proposed algorithms have similar performance to the Mix-ESPRIT algorithm and the Mix-PM algorithm,while have lower computational load.(3)The DOA estimation algorithms based on spectral peak search for circular and non-circular mixed signals with uniform linear array are studied,including Mix-Capon algorithm and Mix-MUSIC algorithm.The proposed algorithm utilizes the signal conjugate form for data expansion.The noncircular characteristics in the extending signal can improve DOA estimation performance.Using the non-circular characteristic,the spectral peak search function of the circular signal and the spectral peak search function of the mixed signal are obtained respectively inspired by the classical peak search method,the circular and non-circular signals can be distinguished through peak contrast in these spectral functions.The proposed algorithms have no need for the estimation of the non-circular phase,and have better angle estimation performance.In addition,when the number of incident signals and the number of receiving array elements are the same,the signal incident angle can be also effectively estimated.(4)The DOA estimation algorithms based on spectral peak search for circular and non-circular mixed signals with two-dimensional(2D)uniform planar array are studied,including 2D-Mix-MUSIC algorithm and RAnk REduced Mixed signals MUltiple SIgnal Classification(RARE-Mix-MUSIC)algorithm.Firstly,the 2D-Mix-MUSIC algorithm under uniform planar array is extended by the MixMUSIC algorithm in the uniform line array.The algorithm has good angle estimation performance and does not need to estimate the non-circular phase.In addition,the incident angles appear in pairs and automatically match,and the circular and non-circular signals can be distinguished,but the algorithm has a high complexity in the two-dimensional peak search process.In order to avoid two-dimensional spectral peak search,the RARE-Mix-MUSIC algorithm using the reduced rank method is proposed.The algorithm separates the incident angles from each other and constructs a spectral peak search function using the separated matrix,so that the two-dimensional spectral peak search is reduced to onedimensional spectral peak search,so only twice one-dimensional spectral peak searches is needed.The estimated value of the signal DOA greatly reduces the complexity of the algorithm while ensuring the performance of the algorithm.