Prediction based Event Triggered Estimation for Networked Systems

This work focuses on development of prediction-based event -triggered estimator for the networked system (NS) with packet delay and packet loss in sensor to estimator network channel. An event-based triggering mechanism limits the number of data transmitted through the network channel. The stochastic model of the sensor to estimator channel is developed using Bernoulli distributed random variables. The one-step predicted measurement is used by an estimator when the data packet is lost or when the event is not triggered. An augmented NS model with the one-step ahead predicted information is developed and the recursive estimator is designed using the orthogonal projection principle. The steady-state property of the proposed event-triggered based estimator is analyzed by showing the trace of prediction error covariance matrix that converges to a unique positive semi-definite value with the given initial conditions. The designed estimator is validated on a mass-spring-damper system model. It is inferred from the simulation results that the performance of the proposed estimator is satisfactory. Copyright (C) 2021 The Authors.