| Airborne phased array radar receives strong ground clutter when operating in the air-search or ground-search mode.The weak targets are easy to be submersed so that the detection performance of radar decreases.Space-time adaptive processing(STAP)is an efficient technique for clutter suppression since it can filter the two-dimensional samples adaptively,which is beneficial to airborne radar target detection,especially for slow targets and small targets.However,the performance of STAP is related to the precise of estimated clutter covariance matrix in practice.In fact,for the non-sidelooking airborne radar,the estimated sample covariance matrix using adjacent samples departs from the true one due to the range-dependent of clutter.Thus the performance of STAP is declined in this case.Focusing on the suppression of clutter for non-sidelooking airborne radar,some preprocessing of clutter samples into homogeneity before STAP is necessary.To achieve a reliable detection of moving targets,the other problem needs to be solved must be interference suppression.Adaptive beamforming can effectively suppress the interference and noise,and protect desired signal simultaneously.Minimum variance distortionless response(MVDR)beamformer is widely used in practice.The beamformer weight vector is calculated by maximizing the output signal-to-interference-plus-noise ratio(SINR)which is related to the interference plus noise covariance matrix(IPNCM)and the steering vector of desired signal.The ideal IPNCM is unavailable and often replaced by the sample covariance matrix.It is necessary to propose robust adaptive beamforming against the mismatches of IPNCM and steering vector of the desired signal,which degrades the performance of MVDR beamformer.With the development of military activities and technology,the level of demand to radar contains not only target detection but also parameter estimation.Target direction-of-arrival(DOA)estimation is of great importance in parameter estimation.In recent decades,one proposes a series of DOA estimation methods,including multiple signal classification(MUSIC),maximum likelihood(ML),and so on.However,to improve the estimated accuracy and resolution capability of target DOA under non-ideal conditions are key techniques in this field.The improvement of STAP performance in nonhomogeneous environments and that of adaptive beamforming and target DOA estimation in non-ideal conditions are considered in this dissertation.The main work of this dissertation is as follows:1.The problem of range-dependent clutter suppression in non-sidelooking airborne radar is studied.The range-dependent of clutter in non-sidelooking airborne radar result in that the estimated covariance matrix departs from the true one,and the performance of STAP degrades.The registration-based compensation(RBC)gets clutter spectrum by temporal smoothing,and then the accuracy of clutter spectrum is limited to the temporal aperture.According to the sparsity of clutter spectrum,we calculate it in each sample based on iterative adaptive algorithm(IAA).And then the clutter covariance matrix can be reconstructed using the compensated data.Thus,a more precise clutter spectrum can be obtained because the aperture loss is not exists.Focusing on the suppression of clutter for non-sidelooking airborne radar,this preprocessing of clutter samples into homogeneity before STAP is efficient.2.The problem of adaptive beamforming against the mismatch of desired signal steering vector and the insufficient of sample is studied.Adaptive beamforming is sensitive to the mismatch of steering vector,especially when the desired signal is contained in the training samples.In view of this problem,a robust adaptive beamforming algorithm via joint estimates of interference plus noise covariance matrix and steering vector is proposed.Firstly,using Capon method to estimate the DOAs of all interferences and signal,which unite with the knowledge of array geometry to form a set of interference basis vectors;then,the basis coefficients can be estimated from LS problem,and a more precise interference-plus-noise covariance matrix can be reconstructed;Finally,we unite the estimate of the DOA of signal with the knowledge of array geometry to form a set of signal basis vectors and estimate the steering vector of desired signal based on the subspace projection.This method can obtain a good performance in small sample size,and is robust to the mismatch of steering vector.3.The problem of target DOA estimation in airborne phase array scanning radar is studied.In order to avoid straddling loss,airborne phased array scanning radar transmits beams with small step angle,and then the same target can be detected again and again in multiple adjacent beams through azimuth scanning.Here,a direction-of-arrival(DOA)estimation algorithm that utilizes multiple beams to obtain an accurate target angle is proposed.The improved target DOA can be obtained by jointly exploiting multiple beams received data due to the fact that the signal energy is partially integrated over multiple different beams.4.The problem of target direction-of-arrival estimation in small sample size and low SNR value is studied.The realization of MUSIC algorithm is often based on the accuracy of estimated noise subspace.For case with small sample,the estimated covariance matrix deviates from the true one due to the cross-term of signal and noise.Besides,for case with low SNR,noise subspace cannot be separated from the signal subspace effectively.Thus,the noise subspace obtained directly from sample covariance matrix is imprecise,which result in performance breakdown.Here,we propose an iterative algorithm to improve the MUSIC DOA estimation performance by modifying the noise projection matrix.Simulation results demonstrate that the proposed algorithm based on the new noise subspace outperforms the classic algorithm,especially in small sample size and low SNR value.5.The fast implementation of beamspace root-MUSIC DOA estimation is studied.We divide the in-band angular sector into multiple small sectors.An arbitrary steering vector in a small angular sector can be well represented by a low-order real polynomial vector using the low-rank properties of steering matrix.Then the problem of beamspace root-MUSIC DOA estimation can be translated into a multi-taper real polynomial rooting problem.The results show that the developed algorithm provides a lower computation complexity and approximately identical performance. |
PDF Full Text Request