GTD Model Parameters Estimation Based on Improved LS-ESPRIT Algorithm

Abstract

With the rapid development of stealth technology, the radar target edge diffraction has become the main scattering source and it has replaced the mirror diffraction. As the core theory of the radar target edge diffraction, the geometric theory of diffraction (GTD) model can characterize the scattering features of radar targets more accurately. So, it is significant to estimate the parameters of the GTD model. Least square-estimating signal parameters via rotational invariance techniques (LS-ESPRIT) algorithm is a classical method to estimate parameters of GTD model. However, the performance of the basic LS-ESPRIT algorithm drops dramatically when the signal-to-noise-ratio (SNR) is low. To tackle this problem, an improved LS-ESPRIT algorithm, which combines a nuclear norm convex optimization method with the basic LS-ESPRIT algorithm, is proposed in this paper. A less noise-sensitive Hankel matrix with low rank can be obtained using the nuclear norm convex optimization method. Combining the new rearranged Hankel matrix with the basic LS-ESPRIT algorithm, positions parameters, type parameters and intensity parameters can be estimated more accurately. To validate the performance of the improved LS-ESPRIT algorithm further, a RCS reconstruction simulation is performed based on the estimated parameters and GTD model. Simulation results show that the estimation accuracy of the improved LS-ESPRIT algorithm is higher at a low SNR. The maximum root mean square error (RMSE) of the position parameters and the intensity parameters are less than 0.005 m and 0.2 respectively. And the accuracy of scattering types can be increased by 70% at most. Moreover, compared with the basic LS-ESPRIT algorithm, the proposed algorithm in this paper has smaller error values of the reconstructed RCS. In conclusion, better estimation, RCS reconstruction property and better stability can be achieved by the improved LS-ESPRIT algorithm.