Self-tuning Decoupled Component Information Fusion Kalman Smoother

For the multisensor systems with unknown noise variances, an on-line noise variance estimator is presented by using a correlation method. According to the ergodicity of the sampled correlation function, it is proved that the estimation of noise variances is consistent. Based on the Riccati equation and optimal fusion rule weighted by scalars for components, a self-tuning decoupled fusion Kalman smoother is presented, which realizes a decoupled fused estimation for state components. By using the dynamic error system analysis (DESA) method, it is proved that the self-tuning fusion Kalman smoother converges to the steady-state optimal fusion Kalman smoother in a realization, so that it has the asymptotic optimality. A simulation example for a tracking system with 3-sensor shows its effectiveness.