Aiming at the defect that the traditional least trimmed squares（ LTS） using hard threshold to determine location and calculate base station

in the non-line of sight（ NLOS） environment

base station with the smallest deviation cannot be separated. A new improved method is proposed. Firstly

the Kalman filtering process is performed on the measured distance

then the base station is orderedly grouped

and the base station set with the smallest residual among all base station combinations is selected as the final position calculation set. Experiments show that the proposed method has improved positioning precision by 24. 1 % and 53. 3 %

compared with the traditional LTS algorithm and least square estimation（ LSE） algorithm in the NLOS environment.