Aiming at the problem that the trajectory prediction error in existing vehicle collision warning system(CWS) is relatively large, a DGPS and other vehicle sensors based vehicle trajectory prediction was proposed. Real time vehicle position estimate was realized by using extended Kalman filter (EKF). Then, constant rate of acceleration change, constant yaw rate acceleration model for vehicle position prediction and V2X technology based cooperative collision warning system(CCWS) were proposed. Based on this, the longitudinal control simulation was realized by using fuzzy theory to validate the method. The simulation results show that the vehicle position prediction model based on constant rate of acceleration change, constant yaw rate acceleration model has lower error, and zCWS can get warnings or brake earlier.