Diffusion-based cooperative space object tracking

Abstract

We propose the diffusion-based enhanced covariance intersection cooperative space object tracking (DeCiSpOT) filter. The main advantage of the proposed DeCiSpOT algorithm is that it can balance the computational complexity and communication requirements between different sensors as well as improve track accuracy when measurements do not exist or are of low accuracy. Instead of using the standard covariance intersection in the diffusion step, the enhanced diffusion strategy integrates the 0-1 weighting covariance intersection strategy and the iterative covariance intersection strategy. The proposed DeCiSpOT algorithm also uses the global nearest neighbor and probabilistic data association for multiple space object tracking. Two typical scenarios including cooperative single and multiple space object tracking are used to demonstrate the performance of the proposed DeCiSpOT filter. Using simulated ground-based electro-optical (EO) measurements for multiple resident space objects and multiple distributed EO sensors, the DeCiSpOT archived results comparable to an optimal centralized approach. The results demonstrate that the DeCiSpOT is effective for space object tracking problem with results close to the optimal centralized cubature Kalman filter.