Based on the practical characteristics of the resources for missile defense early warning system, the muti-agent technology is used to study the problem of how to schedule resource of missile defense ear1y warning system.And two agents which are resource management agent and resource agent are designed. Then, a method of fit value computation of a scheme is given, and the optimization scheme has the maximum fit value. For reducing the time of creating scheme, an improved particle swarm optimization algorithm—self adaptive probability particle swarm optimization (SAPPSO) algorithm is designed, in this algorithm, every element in a particle will be changed by the probability which decided by the fit value of scheduling scheme, this course shows how the particles think. By analyzing a real case, and through a contrast with other methods, the result shows that the SAPPSO algorithm can quickly focus on a better value, which can make the scheme meet the operational requirements.