In wireless localization system based on time difference of arrival（ TDOA）

the non-line-of-sight（ NLOS） propagation of signals will lead to the deterioration of positioning performance. In order to improve the robustness of the localization algorithm to NLOS error

considering that the statistical model of NLOS error is difficult to obtain

a localization algorithm based on robust least squares and convex optimization technology is proposed. Under the condition that only the upper bounds of the absolute values of the NLOS errors is known

the algorithm transforms the highly nonlinear time difference localization problem into a robust least square problem.Then the robust least squares problem is transformed into a convex optimization problem by the convex relaxation techniques. The simulation results show that the proposed algorithm can effectively suppress the influences of NLOS errors and possesses superior robustness compared with the traditional algorithms.