Gaussian approximation in recursive estimation of multiple states of nonlinear wiener systems

Abstract The paper concerns recursive state estimation in Wiener systems where the output is a scalar weighted sum of multiple states. A Gaussian approximation to the computed conditional probability distribution for the scalar sum of states is shown to lead to a practicable numerical algorithm for estimating the multiple states. This algorithm mimics the Kalman filter. Simulations show the method giving good performance when the nonlinearity is severe quantization.