Air collision warning is an important task to ensure flight safety and by the increasing throughput of airports, the air route is congested and high performance air collision avoidance alert is required. ADS-B and airborne GPS receiver is used to obtain the relative position information. With EPU data, from GPS message, ADS-B and TCAS data fuss to obtain an optimal linear fusion and the final filter output shall greatly improve the precision of relative position which is able to provide a more accurate judging condition for air collision avoidance strategy. The simulation experiment results show that the fusion algorithm well performed in computing speed, accuracy, robustness thus significantly improve the ability of air collision alert.