Research On GMTI For Airborne/Spaceborne Side-looking Multi-channel Radar System

This doctoral dissertation addresses the key techniques, such as phase center estimation, channel gain and phase response error calibration, clutter rejection and reduced-dimension processing, moving targets radial velocity estimation and location, targets re-focusing, for ground moving target indication with side-looking multi-channel radar systems. The main topics of this dissertation are listed as follows:1. For the purpose of phase center estimation and channel manifold calibration, a new algorithm which explores phase and amplitude relationshipe among along-track Doppler beam sharping (DBS) images is proposed. The channel manifold calibration procedure uses a power-based criterion to reject a weak clutter patch and/or target for covariance matrix estimation, on the other hand, a phase-based criterion to alleviate the strong target signal contamination problem. Thus the problem of target signal contamination can be alleviated observably. Finally, phase center among multi-channel are estimated based on fitting the linear relationship between unwrap phase of the clutter space steering vector and Doppler frequency adopting the total least square algorithm. The effectiveness for channel mismatch and phase center estimation where a millimeter estimation result can be achieved and the robustness for target signal contamination are confirmed by performance analysis and experimental data..2. Clutter rejection is the prime problem for ground moving target indication (GMTI). We focus on clutter inner undulating which influences the performance of moving target detection. A new algorithm based on generalized noise eigen-subspace which is in existence and orthogonal to clutter subspace is presented. The robustness for clutter inner undulating and synthetic aperture radar (SAR) images registration error is proved by subspace analysis and extensive simulation.3. A new reduced-dimensional method based on joint pixels sum-difference data for clutter rejection and GMTI is proposed. The reduced-dimensional joint pixels sum-difference data are obtained by the orthogonal projection of the joint pixels data of different synthetic aperture radar images generated by a multi-satellite radar system. In the sense of statistic expectation, the joint pixels sum-difference data contains the common and different information among SAR images. Then the objective of clutter cancellation and GMTI can be achieved by adaptive processing. Simulation results demonstrate that the performance loss induced by this method for homogeneous terrain with no co-registration error can be acceptable and better result can be achieved adopting this method in the case of heterogeneous terrain with a finite co-registration error.4. For the purpose of overcoming the contradiction between minimum detected velocity and maximum unambiguous velocity, a new target radial velocity estimation approach which is performed in tow stages is proposed. The coarse estimation procedure uses a track-based criterion to avert the unambiguous problem. Using modified single-snapshot multiple direction of arrival estimation method, a fine estimation of target radial velocity can be achieved in the signal fitting sense. Note that the last stage is carried out at determinate range-Doppler test cell to locate moving target by azimuth searching for that one fitting best to the moving target signal, thus the location performance would not be sacrificed in order to suppress clutter and/or interference and the high resolution radial velocity estimation can be achieved. Therefore, the proposed algorithm is computationally inexpensive. A preliminary result against an airborne experimental data demonstrates the effectiveness of the proposed method. Moreover, the moving target echoes is similar with linear frequency modulation (LFM) signal, so target moving parameters can be deduced by estimating chirp rate and initial frequency. We propose a method which using spatial-frequency joined weight WVD for cross-product terms suppressing at first, and taking edge detecting in image domain followed by applying Hough transformation for line detecting and eliminating cross-product terms at the most, then giving the target moving parameter estimation by using least square algorithm. Advantages of this approach are that cross-product terms are suppressed and accurate chirp rate estimates is obtained.