Interactive multiple models algorithm assumes that the switching model system is Markov. The basic idea of interacting multiple model algorithm is in every moment, suppose a model in the present moment effective condition, through a moment before mixing all filter state estimation of initial conditions to obtain filter matched with this particular model. Interactive multiple model Kalman filter has been successfully applied in multi sensor and multi target tracking. EMD decomposition is linear and complete, and EMD decomposition is superior to the traditional method. The paper presents application of interactive multiple models and EMD analysis in signal management of wireless sensor network. Simulation results show that the average performance is better than the traditional algorithm on system.