The paper addresses a fast implementation algorithm about distributed fusion with CPHD filter. An implementation method of distributed fusion based on maximum probability association (MPA), called MPA-DF, is presented. Though the performance of MPA-DF is a little worse than traditional Generalization Covariance Intersection (GCI) distributed fusion method, called GCI-DF, the former needs lower computational burden and better precision than every single sensor. The proposed fusion algorithm is implemented using Gaussian Mixture (GM) approximations and the numerical results show its performance.