Communication rate analysis for event-based state estimation

Abstract

The average communication in a typical event-based state estimation scenario is analyzed analytically and in robot experiments. A sender observes a linear Gaussian process and sporadically transmits mean and variance to a remote estimator according to an event-based protocol. By exploiting the event-based architecture, an expression for the average communication rate is obtained, which is useful for numerical computation. In addition, approximate closed-form expressions for the steady-state rate (given in terms of problem and design parameters) are derived. The analytic results for the communication rate are verified in experiments of a one-dimensional robot link.