In multi-sensor information fusion system, due to the lack of prior knowledge of environment and the restriction of sensor performance, measurement error (noise) will inevitably be introduced in the whole measurement process; In addition, we often do not know the exact number of targets, even if there is only one target (that is, the single target case), due to the interference of clutter, effective measurements may be repeatedly accepted, so we need to establish the corresponding relationship between the target and the measurement through statistical methods. For the multiple targets case, the situation is more complicated: it is impossible to determine whether the measurement is from the interested target, false alarm or other targets. It is because of the uncertainty and randomness in the sensor observation process and multi-target tracking environment, so that the corresponding relationship between the echo measurement and its target source is destroyed.