Study On Robust Methods Of Nonstationary Interferences Suppression For Radar

In the battle field,radar is inevitably faced with complicated interferences.Robust and reliable interference suppression methods can greatly improve the target detection performance in the complicated electromagnetic environment.However,nonstationary interferences have brought new challenges to existing interference suppression methods,which may generated by relative motions between jammer and radar or special transmitting schemes of the jammer,etc.As a result,it is meaningful to develop some robust methods to suppress nonstationary interferences.In this thesis,our work mainly focuses on spatial nonstationary interferences and time-domain transient interferences,which are two typical classes of high-frequency interferences.The whole contents can be summarized into the following four parts:1.We work on the problem of transient interference localization.After analysing the existing methods,to solve the problem that most existing methods cannot provide accurate localization results and measure the localization performance quantitively,the iteratively censored average detector is proposed to achieve interference localization.This method censors the samples that were judged as interference from the process of iteratively weighted summation so that the reliability of the background estimation can be guaranteed.It is proved that the proposed detector can achieve the constant false alarm ratio.Considering the fact that the localization performance of the proposed detector may slightly degrade when transient interferences appear at the beginning and end of the coherent processing interval,we propose to use a forward-backward localization scheme to improve the robustness.Processing results of experimental data and computer simulations show that the proposed method can provide more accurate localization results than that of the existing methods.Moreover,the proposed detector does not add many more computational costs and is proper for practical application.2.Transient interference suppression is one of the emphases of this thesis.With the procedure of clutter suppression,interference localization,interference suppression and data restoration,we propose to use GAPES(Gapped Amplitude and Phase EStimation),a nonparametric interpolation method,to achieve data restoration instead of conventional linear prediction methods.Some improved parameter setting schemes are discussed.The processing results of the experimental data show that the GAPES method can obtain better performance at high order sea clutter than that of the linear prediction methods,which will help to improve the low-velocity target detection performance.However,GAPES method usually needs several to tens iterations to converge,which means that it will add large computational burdens to the radar system.After finding the reason why the GAPES method needs multiple iterations,we propose to use quadratic smoothing to excise transient interference.Quadratic smoothing can preserve most clutter components while excising transient interferences.In this case,the GAPES method only needs to restore gapped target echoes,which will highly improve its efficiency.The experimental results verify that after quadratic smoothing,the GAPES method only needs one or two iterations to converge.Moreover,we find the reason that the GAPES is still necessary after quadratic smoothing is that the nonadaptive filtering matrix of quadratic smoothing cannot preserve high-velocity target in corruption segments.Based on this analysis,a data-driven adaptive method,named adaptive quadratic smoothing,is proposed.It replaces the conventional filtering matrix in the quadratic smoothing with an adaptive subspace filtering matrix.No more interpolation is needed.The results of experimental data show that the proposed method can achieve the equal performance with that of conventional methods but costs fewer computations.Finally,based on the orthogonality between interference and clutter plus target,we propose to suppress transient interferences and preserve clutter and targets by the subspace estimated from interference-free segments.The main computational costs of this method come from the eigen-decomposition.We analyse the relationship between this method and the adaptive quadratic smoothing and verify the effectiveness of the proposed method with experimental data.3.We try to suppress transient interferences by adaptive beamforming.After analyses,we find that transient interferences suppression can be integrated with the continuous-wave sidelobe interferences and can obtain better performance than that of time-domain methods in the multiple transient interferences case.With the conventional processing procedure of the HF radar,a space-time cascade HF interference suppression procedure is proposed.In this procedure,the interference is first localized.Then the sidelobe transient and continuous-wave interferences will be excised by adaptive beamforming,and the mainlobe transient interferences will be suppressed by time-domain method.Based on the segment processing scheme,we also discuss the processing schemes for the nonstationary transient interferences.The proposed procedure is in accordance with the conventional procedure mostly,which means that it will not add too many extra computational costs to the radar system.Experimental data processing results show the effectiveness and performance superiority over the time-domain methods.Moreover,we want to identify the existence of not only transient interferences but also continuous-wave interferences.Based on the mathematical model of the RFI(Radio Frequency Interference),we propose a RFI suppression method,in which the RFI is localized in every sweep period in the frequency domain and the corresponding secondary data are collected to achieve frequency-domain beamforming.The proposed method can excise both transient and continuous-wave RFIs,and the frequency domain echoes can be obtained directly from the frequency-domain pulse compression.Experimental data processing results show the effectiveness.4.Since null broadening methods do not need to adapt beamforming weights frequently,it becomes a useful method to suppress nonstationary spatial interferences.In practice,adaptive sidelobe cancellation is usually adopted.However,most open literature on null broadening is based on full adaption.To solve this problem,we propose a robust null broadening method that is suitable for sidelobe canceller.This method comes from the classical CMT(Covariance Matrix Taper)theory and the difference lies in that it tapers both the covariance matrix and the mutual correlation vector of the sidelobe canceller.The taper matrix and vector can be calculated off-line with few extra computational costs.Computer simulations verify the effectiveness of the proposed method.