Research On Target Parameter Estimation Approaches For Array Radar Under Nonideal Conditions

Array signal processing mainly studies the data processing of signal echoes received via sensor arrays with different spatial positions to enhance the gain of useful signals and obtain their characteristic information.Compared with the traditional single directional sensor,the array system can provide extreme advantages and benefits in beam steering,signal enhancement,interference suppression and performance improvement of target resolution.Parameter estimation and localization of target signals is a fundamental hot research topic in array signal processing,which is widely used in radar,communications,autonomous vehicles and other fields.The traditional target parameter estimation and localization methods based on subspace and sparsity theory can show excellent signal identification performance in the presence of the array manifold matrix without uncertainties and Gaussian white noise.However,in practical engineering applications,the non-ideality of factors such as external scenes and antenna craft will cause the array steering matrix to deviate from the ideal configuration set in advance,and the target may even have coherence and the noise is extremely likely to be colored noise with unknown statistical properties.Therefore,the effective suppression/eradication of various forms of non-ideal factors becomes an essential research topic for target parameter estimation and its accurate positioning.Moreover,existing radars are facing the prominent problems of diversification of target dimensions and maximization of information utilization.Traditional phased array and frequency scanning antenna can only form angle-dependent beampatterns,and cannot achieve joint parameter estimation and localization of target signals from the range domain.In fact,the target signal can usually be described by parameters such as its spatial position(angle and range),the speed of movement,and the scattering properties of the target.The frequency diversity array can enjoy a range-dependent beampatterns by means of a small frequency offset applied to the transmit antenna,thereby improving the localization capability of the target signal from the range and angle domains.Nevertheless,the coupling of range and angle in the transmit steering vector of FDA radar makes it unable to effectively achieve the joint parameter estimation of target signal.Thus,how to realize the decoupling of range and angle in FDA radar as well as improve the parameter estimation and localization performance of target signal has important scientific research significance and application value.Relying on the Key Project of Basic Strengthening and the Science Foundation for Distinguished Young Scholars of Shaanxi Province,this dissertation conducts research on the key technologies in array channel calibration and target signal parameter estimation and localization.The main content can be summarized as the following five parts:1.Aiming at the performance degradation of traditional target parameter estimation methods in the presence of unknown nonuniform noise and limited snapshots,a reduced signal covariance matrix is constructed based on the rank-one correlation model,and a sparse reconstruction method for target parameter estimation combined with vectorization operation is proposed.Firstly,the proposed approach obtains a reduced signal covariance matrix that can alleviate the influence of nonuniform noise based on an improved Rank-one correlation model.Then,an equivalent signal vector with an extended virtual aperture is obtained by applying the vectorization operation,which enables it to achieve high-resolution estimation of the target signal under limited snapshots with robust noise suppression performance.Furthermore,two efficient matrix recovery algorithms are derived to reconstruct the whole signal covariance matrix from the reduced covariance matrix by exploiting the property that the diagonal elements of the signal covariance matrix are equal in uncorrelated signal scenarios.The first is to calculate the diagonal elements of the signal covariance matrix by using the off-diagonal elements based on the matrix decomposition theory,while the other is to achieve this task by utilizing the low rank characteristic based on matrix completion theory.Therefore,benefiting from the comprehensive utilization of the entire covariance matrix information,the proposed approach can provide excellent target parameter estimation and resolution performance.Simulation results show that the proposed method has superior performance behavior in solving compactly placed target and multi-target scenarios.2.Aiming at the poor angle estimation accuracy and pseudo-peaks of target signals in the presence of nonuniform noise,a sparse representation parameter estimation method based on tail optimization is proposed.This part also considers the target parameter estimation problem in coherent target and gain/phase errors scenarios,respectively.For the case that the target is coherent,the proposed method models the incident signal as its autocorrelation and cross-correlation components based on the least squares criterion,thereby constructing a noise-free covariance matrix composed of diagonal and off-diagonal elements associated with the target waveform covariance matrix.For the one of gain and phase errors between array elements,the proposed one uses parameterized array steering vectors to incorporate unknown again/phase error entries into sparse signal vectors,thereby suppressing the negative effects of array channel errors.Moreover,an equivalent received signal vector similar to that of a single snapshot is obtained by applying the difference coarray principle,and then the nonuniform noise power is removed by leveraging a designed transformation matrix.Finally,a tail optimization based signal sparse reconstruction approach is derived that can significantly suppress false peaks,enabling sharp spatial-domain spectral peak estimation.The advantage of the proposed method is that it can not only enjoy a large virtual array aperture and anti-noise nonuniformity,but also can significantly suppress pseudo-peaks and has high computational efficiency.3.Aiming at the poor accuracy and resolution of coherent target angle estimation in the presence of spatially correlated Gaussian noise,a coherent target parameter estimation method based on improved sparse representation is proposed.Firstly,based on a symmetric uniform linear array,the method utilizes a difference strategy to obtain noise-free elements in the signal covariance matrix,thereby eliminating the influence of spatially correlated Gaussian noise.Subsequently,an equivalent signal vector that can directly ignore the coherence between targets is constructed by applying a square operation to each column of the noise-free covariance matrix.This enables larger array apertures and avoids the aperture loss problem associated with traditional spatial smoothing algorithms used to achieve decoherence of coherent targets.Finally,a weight vector that can significantly enhance the performance of signal sparse reconstruction is devised by taking advantage of the signal subspace theory to improve the parameter estimation performance of target signal.Simulation results show that the proposed method can offer excellent coherent target parameter estimation performance in low SNR scenarios with low computational complexity.4.For the unknown mutual coupling between array elements and nonuniform noise,most of the existing target parameter estimation methods mainly use the prior information of incoherent signals to complete the array channel calibration and target signal localization.Therefore,this kind of algorithm is not suitable for purely coherent target scenarios.Aiming at this problem,two coherent target parameter estimation methods are proposed which can solve unknown mutual coupling and nonuniform noise simultaneously.What the two methods have in common is that they both use the least squares criterion for the signal covariance matrix to suppress noise nonuniformity.The difference is that the first method simultaneously achieves mutual coupling suppression and coherent target decoherence by applying the classical spatial smoothing algorithm to a subarray covariance matrix selected via the middle subarray scheme.The other derives a target parameter estimation method without taking mutual coupling account into based on a designed sub-array intermittent working strategy.This approach can not only comprehensively utilize the target echo information received via the whole sensor,but also enjoy a larger array aperture.In addition,benefiting from the used array shift invariance scheme,the proposed method does not require spectral searching and has high computational efficiency.Simulation results show that the proposed method can effectively overcome the influence of unknown mutual coupling between array elements and noise nonuniformity,and has excellent estimation performance in solving coherent target signals.5.Aiming at the performance degradation of target localization with FDA-MIMO radar in the presence of unknown mutual coupling,a joint range and angle estimation method based on dual subspace and eigenmatrix calibration is proposed.Firstly,DOA estimation can be directly realized by using the constructed two subspaces,which does not require both take into mutual coupling and spectral searching account.Subsequently,a phase registration scheme is devised by applying eigenmatrix matching to decouple the range and angle in the transmit steering vector of FDA radar.Herein,two scenarios where targets have the same DOA and different DOA are investigated,and two efficacious range estimation algorithms are presented successively.For the case of the different DOA,the proposed approach can directly obtain the target range information by exploiting the array shift invariance.While for same DOA situations,a demerger technique based on quadratic eigen-decomposition is introduced to avoid the information fusion problem in range estimation.Furthermore,an effective phase ambiguity resolution method is derived to achieve accurate range information acquisition.The proposed approach can directly complete the joint estimation of both range and angle regardless of the negative influence of mutual coupling between array elements,and does not require additional multi-parameter pairing processing.Simulation results confirm that the proposed method can provide excellent mutual coupling suppression and parameter estimation performance with relatively low computational cost.6.Aiming at the coupling of range and angle in the transmit steering vector in the FDA radar,an automatic decoupling target localization method based on bistatic MIMO and FDAMIMO dual-mode radar is proposed.Firstly,the bistatic MIMO and FDA-MIMO dual-mode radar system model is established,and the DOA estimation of the target signal is achieved by using the joint echo data received via this dual-mode radar.In order to solve the coupling problem of range and DOD in FDA-MIMO radar mode,an effective decoupling method is developed by taking advantage of the monostatic-like characteristics of the dual-mode radar at the transmitter.Specifically,by constraining an eigenmatrix obtained from the received signal of the dual-mode radar to the subspace data corresponding to the MIMO radar and FDA-MIMO radar,respectively,the phases of the transmit steering vectors in different radar modes can be accurately registered.Then,the automatic decoupling of range and DOD in FDA-MIMO radar mode is accomplished by leveraging the phase information of the transmit steering vector obtained via the MIMO radar.This does not require complex array configuration and additional parameter pairing processing.Moreover,a phase ambiguity resolving scheme is devised to achieve accurate range estimation of the target signal.Finally,by using a constructed range compensation factor an enhanced DOD estimation algorithm that can comprehensively utilize the entire transmit steering vector information of the dual-mode radar is derived to attain a significant improvement in the estimation accuracy of the corresponding parameter.The proposed method is extremely effective in reducing the complex array design and high computational complexity caused by attempts to improve the estimation performance of multi-dimensional parameters(range,DOD and DOA)in multi-target scenarios.Simulation results demonstrate that the proposed approach enjoys excellent target parameter estimation and localization performance under design-based MIMO and FDAMIMO dual-mode radar.