This paper presents a distributed estimate and fusion algorithm for the sensor networks whose nodes obey the descriptor system model. Firstly, the singular value decomposition is used to transform the singular system into two equivalent reduced order sub-systems, which yields a set of multi-rate systems by using the multi-rate sampling scheme.
The local Kalman estimators and the filtering error covariance matrices are then obtained by means of the re-organized innovation and the orthogonal projection principle. The fusion estimators are further proposed with a fusion rule weighted by matrices. The algorithm presented in this paper is more precise than the local estimator with the signal sensor. Finally, numerical examples are presented to illustrate the feasibility and effectiveness of the proposed algorithm.