Abstract: A Kalman filter positioning algorithm applied in BeiDou and INS integrated navigation system is proposed. Then, the observability, controllability and stability of integrated navigation filter algorithm is analyzed. The observability of every state is analyzed using method of numerical value analysis. Because the clock-error state can not be filtered, otherwise the algorithm may be divergence, we make improvement on the filter algorithm model, using the change rate of pseudo-ranges and INSs acceleration as observations. Finally, it is explained in emulation mode that our improved integrated navigation algorithm model can improve the running stability of integrated navigation system effectively, its ability of resisting outburst error is enhanced effectively, and the clock-error state can be filtered even when the original location error is up to 13000 meter.