Efficient Parameter Estimation for Cone-Shaped Target Based on Distributed Radar Networks

An echo signal received from a cone-shaped target with micro-motion is composed of a linear mixture of individual signals radiated from multiple effective scatterers with the occlusion effect, resulting in difficulties in parameter estimation for ballistic target discrimination (BTD). To solve this, conventional methods have been based on the sophisticated signal decomposition techniques using a 2D joint time-frequency (JTF) image or a 2D radial-range (RR) history image; however, they are inefficient for real-time BTD due to complex 2D image processing. Therefore, we propose a new parameter estimation framework consisting of five stages: 1) a normalization step; 2) signal decomposition and data association using independent component analysis in the distributed radar network; 3) estimation of dynamic parameters using 1D micro-Doppler frequency trajectories; 4) restoration of 1D RR histories; and 5) estimation of geometric parameters using the restored 1D RR histories. In particular, ICA of stage 2 is more time-saving than the conventional mathematical model-based methods using the 2D JTF image due to signal decomposition using the 1D normalized echo signals. Moreover, in the stage 4, high-quality 1D RR histories can be restored in spite of using the 2D RR history image with low resolution, compared with the conventional methods using 2D RR history image of very high resolution. In the simulations, we observed that our proposed framework is capable of performing efficient parameter estimation for the real-time BTD.