The main contribution of this paper is to effectively track ballistic missile target during its boost phase, by utilizing heterogeneous measurement information from SBIRS and OTHR. In regard of the missile motion properties during its boost phase and the working principle of the sensors, employ the boost phase motion model and the sensor measurement model. Under the UKF filtering framework in nonlinear system, centralization and distribution fusion strategies are proposed for trajectory estimating in boost phase and target tracking. The simulation result shows that the overall data fusion filtering performance from multiple sensors is better than the single sensor filtering performance.