Chandrasekhar-type recursive Wiener estimation technique in linear discrete-time stochastic systems

Abstract

Previously, certain fast algorithm, called Chandrasekhar-type one-step ahead predictor, for recursive least-squares (RLS) estimation by the Kalman one-step ahead predictor is developed in discrete-time systems. In this paper, the Chandrasekhar-type recursive Wiener filter and fixed-point smoother are designed by factorization of increment of the Riccati variable, which is the auto-variance function of the filtering estimate, in the RLS Wiener filtering algorithm in linear discrete-time wide-sense stationary stochastic systems. In general, the characteristic of the Chandrasekhar-type filter is that the filter gain is directly updated recursively in the algorithms. The total number of operations in the new filter algorithm is less than the Riccati-equation based RLS Wiener filter, with significant reductions being obtained.