Smoothing Algorithm for Linear Systems with General  Correlated Measurement Noises
Han Chongzhao¹,Wang Jie¹,Li Xiaorong²
(1. Xi'an Jiaotong University, Xi'an 710049, China; 2. University of New Orleans, USA)

Abstract: Based on the forward and backward filtering estimates a smoothing algorithm is developed for linear systems with general correlated measurement noises by using the linear unbiased minimum variance estimation formula. Because of the correlation of the measurment noises the posterior mean of the noise is not always equal to its prior one and can't be calculated. Hence, the proposed algorithm is suboptimal. When the forward and backward filtering results are known, the proposed algorithm has low computional complexity and can be realized easily. Through a simulation example it is indicated that the result of the proposed smoothing algorithm is better than that of the forward, backward filtering or Kalman smoothing algorithm, where the measurement noises are assumed to be uncorrelated.
Keywords: correlated noise;forward filtering;backward filtering;smoothing;fusion;linear unbiased minimum variance estimation