Study On Moving Target Detection And Localization For Airborne Array Radar

Moving target detection and localization for airborne array radar is an important part of electronic warfare. Its research has been devoted to space-time adaptive processing (STAP) for phased array radar, direction finding of target signal and interferences, and parameter estimation of moving target. Clutter spectrum depends on range for airborne non-sidelooking and bistatic radar system, which degrades the performance of conventional STAP. And the conventional methods for direction-of-arrival (DOA) estimation suffer a low resolution in practice. Recently, multiple-input multiple-output (MIMO) technique is introduced for target detection and localization to combat target fading and increase the degrees of the freedom in adaptive processing. This thesis investigates the challeging problem in the field of target detection and localization for airborne radar. We deal mainly with optimum range-dependence compensation in STAP for a non-sidelooking monstatic radar and bistatic radar system, the DOA estimation in the presence of broadband signal and mutual coupling. Furthermore, some techniques of MIMO radar are also discussed. The main works of this thesis can be summarized in detail as follows.1. Chirp z and resampling methods for the mitigation of the range dependence are proposed in Chapter 2. And the convex optimization is used to design the transmiting and reveiving beam so that clutte can be better suppressed.The range dependency implies that the snapshots used for the estimation of the covariance matrix are not identically distributed for all secondary range bins, which degrades the performance of space-time adaptive processing (STAP) on clutter suppression. According to clutter model, we propose two compensation methods: Chirp z-transform (CZT) and space-time resampling. Exploiting the scale characteristic of the clutter spectrums at any two range bins, the CZT method can stretch clutter spectrum with a unique scale at each range bin in the Doppler-angle domain, resulting in an identical clutter Fourier spectrum in both mainlobe and sidelobes for all range bins. This method can be realized by 2-D CZT with low complexity. After IFFT of the obtained spectrum, the clutter covariance matrix at the range bin under test can be better estimated. The space-time resampling method can mitigate the range dependence of clutter spectrums by adjusting the space-time sampling rate. Different from other compensation methods, the method uses interpolation and filtering to obtain an identical clutter spectrum in both mainlobe and sidelobes. STAP is then applied to the output of the filter. Simulation results show the validity of the proposed two methods.When using airborne forward looking planar antenna to detect ground moving target, target signal may be masked by strong clutter due to high sidelobe level of the antenna pattern. Only the receiving beam is designed to reject clutter and noise in the conventional STAP. In some known envirment, we can use electronic maps, SAR images and GPS to obtain the knowledge of the detected scenario. We synthesis transmiting beampattern via convex optimization to allow target signals to better compete with strong ground clutter returns. In the azimuth, utilizing the obtained knowledge, transmiting beampattern can be synthesized with some notches in some given directions where strong clutter and interferences exist. And transmiting beampattern has a low sidelobe illuminating short ranges and a high sidelobe focused into sky and remote ranges, which can greatly reject clutter in the elevation. With insufficient training data due to a dispersion of clutter spectrum along range, adaptive beampattern with low sidelobes can be obtained by convex optimization when detecting remote targets.2. Chapter 3 investigates the range dependence and its mitigation methods for four typical bistatic radar systems. The property of clutter for the bistatic radar system is more complex than that for the monostatic one. The thesis presents the relationships between clutter Doppler frequency and range for the side/forward looking radar system with the vertical/horizontal baseline. Theoretical analysis and simulations show that the linear relationship between Doppler and direction at every range bin for the sidelooking/vertical system. For all range bins, the centers of clutter spectrums fit well, and the further the spectrum from the center, the worse the fitness. The side looking/ horizontal radar system has a wider clutter spectrum region in the mainlobe than the vertical baseline. The clutter bandwidth depends greatly on the orientation of receiving array axis relative to the aircraft velocity vector for the two types of forward looking bistatic system.Based on this, a spectrum-rotation method for the range-dependence problem is provided for the sidelooking/vertical bistatic system. The method applies space-time rotation matrix to the snapshots. As a result, the spectrum at each range bin can be rotated with the corresponding angle to realign the reference spectrum. For other three bistatic systems, according to the given relationships between Doppler frequency and range, clutter spectrums are compensated for by the well known angle-Doppler compensation approach. 3. A scale method for wideband DOA estimation and a method based on singular value decomposition (SVD) in the presence of mutual coupling are proposed in Chapter 4. DOA estimation is important for airborne radar. In practice, some factors, such as wideband source and array mutual coupling, degrade the performance of the conventional DOA methods. A scale method for DOA estimation of wideband signals received by an uniform linear array antenna is proposed. The method does not require the preliminary DOA estimates. Due to the fact that there is only a scale expansion difference between the spatial Fourier spectrums of any two narrowband components, with the use of CZT, the spatial Fourier spectrum at each frequency component can be stretched with a unique scale to be identical to the reference spectrum, thus achieving a focused covariance matrix. The method has further low computational complexity via the fast realization of CZT.A method for direction finding in the presence of mutual coupling is presented. The method uses the SVD to calculate coupling coefficients from the initial estimates of the direction-of-arrival (DOA). Exploiting these calculated coupling coefficients, the DOA can be accurately estimated by the multiple signal classification (MUSIC) algorithm. Based on this, Blind calibration is realized by only several iterations between coupling coefficients and DOA estimation. To deal with the local optimum problem in the iterations, the improved methods in the presence of single and multiple sources are presented respectively.4. Methods for target localization and velocity estimation in statistic MIMO radar are provided in Chapter 5. And the performance of coherent radar is examined.MIMO radar is diveided in statistic MIMO and coherent MIMO. For statistic MIMO radars, AOA (angle of arrival) position estimation by the least squares algorithm offers an undesirable performance. Our analysis shows that this results from the fact that the coefficient matrix has errors. And a method based on the constrained total least squares algorithm, exploiting the correlation among coefficient matrix errors, is proposed. The mean-squared-error (MSE) of the estimated position is given.Based on statistic MIMO model, the angle and velocity of moving target are estimated by the relationship among target velocity vectors for all transmitter-receiver pairs. Further, radial velocity relative to the receiver is computed by projecting velocity vector onto the looking direction (from receiver to target), and its performance analysis is done. Theoretical analysis and Simulation results show that the velocity estimation performance for statistic MIMO radar depends on the angle between any two vectors from the target to the receiver and transmitters, and the greater the angle, the smaller the estimation error.Coherent MIMO radars have multiple spatially separated transmitters and co-located receivers. MIMO beam performance is investigated and compared with the two-way beam (the receiving beam multiplied by the transmiting one) of the phased array radar. Theoretical analysis indicates that the full MIMO beam that can utilize all virtual array elements is identical to the two-way beam of the phased array radar; the partial MIMO beam which exloits only those elements with different phase centers is narrower, but suffers a gain loss; the partial MIMO may avoid aliasing in space even when the transmiting array elements are spaced at more than half wavelength; as a scanning radar, the full MIMO radar provides the same detection range as the phased array radar, and the shorter detection range for the partial MIMO radar.