The decentralized information fusion algorithms for multiple heterogeneous sensors platforms with different types and dimensions was developed to meet the needs of the target tracking with active and passive sensors. The decentralized information fusion algorithm was scalable, heterogeneous and reconfigurable. The performance cost function and constraints model of communication, collision avoiding and control for decentralized optimal control of multiple heterogeneous unmanned aerial vehicles(UAV) in cooperative target tracking were established to maximize the local information entropy obtained by information fusion and the quality of information observed by each UAV. The cooperative target tracking based on multiple heterogeneous UAV was implemented using decentralized model predictive control. The effects of imperfect communication on decentralized information fusion and cooperative control were investigated.