Self-tuning reduced dimension weighted fusion white noise estimator with input estimation

For the multisensor linear discrete time-invariant systems with unknown constant input and unknown noise statistics, the on-line noise statistic estimator based on the improved correlation function method, and the on-line input estimator based on the RELS algorithm are presented. Using the Kalman filtering method, a self-tuning reduced dimension weighted measurement fusion white noise deconvolution is presented based on the self-tuning Riccati equation. Based on the dynamic error system analysis method, its asymptotic optimality is proved, i.e. it converges to the optimal fusion steady-state white noise estimator in a realization. A simulation example for a 3-sensor system with Bernoulli-Gaussian input white noise shows its effectiveness.