Abstract: The positioning accuracy of traditional multilateration method(MLAT)in airport surface surveillance is easily affected by the observation error in NLOS environment.To solve this problem,a high-precision MLAT method combining M-estimation and extended Kalman filter(EKF)is proposed.The TDOA data measured by the surface receiving station was constructed into a numerical model.Using the idea of Huber-M estimation,the observation updating step in standard EKF was changed to a weighted least square linear regression problem,so as to improve the anti-interference ability of EKF to non-Gaussian observation noise.The improved EKF was applied to location estimation.The simulation results show that the proposed method is robust to the observation noise of TDOA and achieves high positioning accuracy.