Centralized Robust Multi-Sensor Chandrasekhar-Type Recursive Least-Squares Wiener Filter in Linear Discrete-Time Stochastic Systems with Uncertain Parameters

In the centralized robust multi-sensor recursive least-square (RLS) Wiener filtering algorithm, the number of recursive equations increases compared to that of the centralized multi-sensor RLS Wiener filter in linear discrete-time stationary stochastic systems with uncertain parameters. Due to the increase in the number of recursive Riccati-type algebraic equations, the accumulation of round-off errors is not negligible. The round-off errors cause unstable numerical characteristics of the filter, especially for the small variance of the observation noise. To reduce the round-off errors as the first attempt in the research field of centralized robust multi-sensor estimation this paper designs the Chandrasekhar-type centralized robust multi-sensor RLS Wiener filter, which updates the filter gains recursively. To verify the effectiveness of the proposed filter, a numerical simulation example is demonstrated and its estimation accuracy is compared with the centralized robust multi-sensor RLS Wiener filter and the centralized multi-sensor RLS-Wiener filter. The obtained results show that the proposed filter exhibits better stability. Keywords— Chandrasekhar-type centralized robust RLS Wiener filter; Multi-sensor information fusion; Base station; Autoregressive model; Uncertain stochastic systems.