Numerical Simulation of Robust Recursive Least-Squares Wiener Estimators for Observations with Random Delays and Packet Dropouts in Systems with Uncertainties

Abstract

Seiichi Nakamori

This paper investigates the numerical estimation characteristics of the robust recursive least-squares (RLS) Wiener estimators by using the observed values with random delays, packet dropouts, and out-of-order packets for the systems with or without the uncertain parameters in the system matrix and the observation vector. The estimation characteristics are compared with the existing estimators. (1) The estimation accuracy of the robust RLS Wiener filter is superior to the RLS Wiener filter and fixedpoint smoother. (2) The estimation accuracy of the robust RLS Wiener filter is superior to the RLS Wiener filter and fixedpoint smoother, which are designed for the delayed and uncertain observations, except for the observation noise N(0, 0.5 2 ), provided that the signal exists in the observed values. (3) In the case of the observations with random delays and without including the uncertain parameters in the system matrix and the observation vector, the estimation accuracies of the robust RLS Wiener filter and fixed-point smoother are superior to the RLS Wiener filter and fixed-point smoother, which are designed for the delayed and uncertain observations. It should be noted that the robust RLS Wiener estimators do not assume any knowledges of the probabilities of the random delays, and the uncertain parameters.

PDF

Share this article

Get the App