Adaptive Motion Model Selection for Direction-of-Arrival Estimation in Radar Space Surveillance Using Constrained Admissible Regions

An increase of objects in space leads to the need to establish Space Situational Awareness (SSA). An essential part of it is space surveillance using i.e. radar. One part of radar signal processing is the Direction-of-Arrival (DoA) estimation. Within this topic a result of prior work was, that the DoA estimation needs to be combined with a time dependent component to compensate the high velocity of objects in space. We further analyze this problem considering the orbital dynamics as well as different Signal to Noise Ratio (SNR) scenarios. We show that the compensation of the DoA motion might lead to worse results compared to a static DoA assumption. Consequently, the main contribution of this work is an adaptive motion model selection method. An a prioi detection in Range-Doppler-Acceleration (RDA) domain is used to build up a Constrained Admissible Region (CAR), which is used with the corresponding detection SNR to evaluate different error bounds and might find the best fitting motion model. We provide a typical numerical example of a Sentinel 3B satellite pass and show that the proposed method has advantages with respect to the Root Mean Squared Error (RMSE) compared to a single motion model DoA estimation.