Efficient implementation of maximization likelihood estimation to constrained measurement random latency probability in nonlinear system

Abstract

This paper focuses on quickly identifying the unknown or time-varying random latency probability (RLP) of the measurements in the nonlinear networked multi-sensor system by resorting to the efficient implementation of maximization likelihood (ML) estimation. Firstly, the full-probability likelihood computation is equivalently transformed into a log-likelihood function summation form parameterized by RLP through Bayes' rule. Secondly, the computation of the log-likelihood function is further transferred by skillfully introducing Jensen's inequality for facilitating the rapid maximization. Thirdly, the simple identification result of RLP is obtained by constructing Lagrange operator to maximize the transferred log-likelihood with the RLP parameter constraint. Finally, an example motivated by the maneuvering target tracking application is presented to demonstrate the superiority of the new method.