Distributed Fusion Estimation for Nonlinear Binary Sensor Systems

This paper investigates the problem of distributed nonlinear fusion estimation for clustered binary sensor networks. In this framework, each cluster head is responsible for performing local estimation and communicating with other cluster heads to obtain a unified fusion estimation. By using prediction measurement to construct a threshold model with uncertainties, and then a local cubature kalman filter based estimator is designed, where the model uncertainties were eliminated by matrix inequality technique. Then, a distributed fusion estimation algorithm based on enhanced sequential covariance intersection fusion criterion independent of fusion order is proposed. Meanwhile, the fusion results can be fed back to the local estimator to further improve its performance. Finally, a damped mass-spring system is employed to demonstrate the effectiveness of the proposed method.