Suboptimal continuous filtering based on the decomposition of observation vector

Abstract This paper presents suboptimal linear and nonlinear filtering methods based on observation vector partition. The usual Kalman filter and extended Kalman filter are decomposed into separate local filters by observation vector partition method. The proposed filtering methods allow parallel processing of information and reduce both off-line and on-line computational requirements (particularly in dynamic systems having a large dimensional observation vector). Examples demonstrate the efficiency of the proposed suboptimal filter.