Research On Non-homogeneity Detection And Clutter Suppression For Airborne Radar

Airborne radars offer the potential for a significant improvement in target detection performance over conventional ground based systems. This is primarily due to the mitigation of terrain masking effects associated with the low altitude siting of ground radars. In airborne radars, clutter suppression becomes arduous due to several effects including: an increase in clutter backscatter at high grazing angles; clutter spectral spreading due to platform motion; and wide variation in ground clutter returns over the surveillance volume. It is this spatial variation or non-homogeneous nature of the clutter environment which presently limits the detection performance of STAP-based airborne radar.The basic theory of STAP is relatively complete. However, in practical engineering applications, STAP is still faced with many problems. STAP for low sample support applications and STAP in non-homogeneous environments are hot and urgent problems in the field of STAP research. This paper is focused on studying the above two aspects, and the main contents are organized as follows:1.The performance improvement of STAP methods in a limited sample support environment is studied. When airborne radar operating in non-homogeneous environments, the independent and identically distributed training samples obtained by STAP are always limited, which results in an inaccurate estimation of the clutter covariance matrix and a severely degraded STAP performance. To solve this problem, a conjugate gradient method based on the forward and backward averaging is proposed. The results show that the method can improve the radar moving target detection performance in a limited sample support environment.2.The outlier detection problem is studied. One source of outlier data is the targets themselves. For example, if one is trying to detect an individual target adaptively in a dense target environment(such as an airborne formation), the other target returns located in distinct range cells about the individual target with essentially the same velocity vector can be present in the training data. All of the targets have approximately the same steering vector. The presence of the other target signals corrupts the covariance estimation severely,resulting in performance degradation of the conventional non-homogeneity detector. To address the problem, a non-homogeneity detector based on unequally weighted generalized inner product is developed. The deleterious effect of outliers on the covariance calculation is eliminated by unequally weighted scheme to the training data set. Simulated and measured data validates that the proposed method can effectively remove outliers from the training data and improves the robustness of the traditional generalized inner product detector.