Fusion estimation for nonlinear multi-sensor networked systems with packet loss compensation and correlated noises

Purpose: This paper aims to give the centralized and distributed fusion estimator for nonlinear multi-sensor networked systems with packet loss compensation and correlated noises and give the corresponding square-root cubature Kalman filter. Design/methodology/approach: Based on the Gaussian approximation recursive filter framework, the authors derive the centralized fusion filter and using the projection theorem, the authors derive the centralized fusion smoother. Then, based on the fast batch covariance intersection fusion algorithm, the authors give the corresponding results for distributed fusion estimators. Findings: Designing the fusion estimators for nonlinear multi-sensor networked systems with packet loss compensation and correlated noises is necessary. It is useful for general nonlinear systems. Originality/value: Throughout the whole study, the main highlights of this paper are as follows: packet loss compensation and correlated noises are considered in nonlinear multi-sensor networked systems. There are no relevant conclusions in the existing literature; centralized and distributed fusion estimators are derived based on the above system; for the posterior covariance with compensation factor and correlated noises, a new square-root factor of the error covariance is derived; and the new square-root factor of the error covariance is used to replace the numerical implementation of the covariance in cubature Kalman filter (CKF), which simplified the problem in calculating the posterior covariance in CKF.