Abstract: The positioning time-consuming of the differential global navigation satellite system (DGNSS) real time kinematic (RTK) technology almost depends on the Kalman filter implementation. There is a mass of matrix operation in Kalman filter implementation, so matrix operation optimization is an effective mean to reduce time-consuming of RTK. The method in this paper improves the matrix operation of Kalman state variable update and it’s covariance update based on the special Kalman state transfer matrix in RTK positioning algorithm. The results of motion platforms test show that the improved RTK positioning timeconsuming is about 7 times less than before.