This paper is devoted to global positioning system (GPS)-based fault-tolerant satellite orbital states estimation which is used for the pseudo-ranging model. GPS is widely used by many researchers to determine the orbit state vectors of satellites, but errors in GPS measurements can cause major errors in satellite location detection. Therefore, there is a need for an architecture that will resist possible measurement errors and GPS malfunctions. This study aims to present the novel approaches to tolerate the errors that may take place in GPS measurements. The satellite localisation has been detected via a pseudo-ranging model between four GPS satellites and the target satellite. The Keplerian equations and orbit elements are used for the satellite dynamic model with orbital J2 perturbation which is due to the flattened shape of the earth is taken into account via the extended Kalman filter (EKF) algorithms. The reconfigurable and the adaptive version of fault-tolerant localisation architecture eliminate the GPS measurement errors during estimating satellite position vectors. The comparison of the prediction architectures of the fault tolerant satellite state vectors originally proposed. At the same time, it is possible to use fault tolerant architectures designed to detect whether GPS satellites are malfunctioning or not.