In the parameter learning step, sight conditions are updated according to
the measurement and the estimation and mobile state. Simulation results show that this method could achieve a good tracking performance and the NLOS parameters can be effectively inferred.