Abstract
The precision of space registration is generally low with the situation of a big random error in radar network. In the case of neglecting model linearization errors, this shortcoming is due to the fact that the coefficient matrix of measurement equation exists errors. This paper presents the constrictive total least squares algorithm based on the Earth-centered Earth-fixed (ECEF) coordinates, which exploits the correlation among coefficient matrix errors. Simulation results show that the method has higher registration accuracy and faster convergence speed, compared with least squares and generalized least squares algorithm.