Abstract: There are three basic fusion algorithms for target-tracking, which are centralized fusion algorithm, distributed fusion algorithm and hybrid fusion algorithm. Centralized fusion algorithm can achieve the highest tracking accuracy but it needs heavy processing load in fusion center and higher communication load. Recently, the distributed algorithm has received significant attention in multi-sensor target tracking for its light processing load in fusion center and lower communication load. In linear systems the distributed fusion can succeed to reconstruct the optimal centralized fusion estimate by combining the local estimates. In nonlinear systems, converted measurement Kalman filtering algorithm (CMKFA) is better than extended Kalman filtering algorithm (EKFA) for target tracking. This paper mainly studies data fusion algorithm based on converted measurement Kalman filter (CMKF) for target tracking in nonlinear systems. From theoretical analysis, it is derived that the distributed converted measurement Kalman filtering algorithm (DCMKFA) can basically reconstruct centralized fusion estimate. And simulation results can prove this conclusion. So DCMKFA is a better distributed fusion algorithm in nonlinear systems.

Key words: tar get t racking, data fusion, centr alized fusion algorithm, distr ibuted fusion algor ithm