Sequential detection of Markov targets with trajectory estimation

TL;DR

This paper introduces a sequential detection and estimation method for Markov targets using a combined SPRT and MAP approach, with theoretical analysis and application to radar surveillance.

Contribution

It proposes a novel sequential detection and trajectory estimation procedure for Markov systems, with asymptotic optimality analysis and practical radar application.

Findings

Asymptotic optimality conditions are established for the proposed test.

The method effectively detects and estimates Markov target trajectories.

Application to radar surveillance demonstrates practical utility.

Abstract

The problem of detection and possible estimation of a signal generated by a dynamic system when a variable number of noisy measurements can be taken is here considered. Assuming a Markov evolution of the system (in particular, the pair signal-observation forms a hidden Markov model), a sequential procedure is proposed, wherein the detection part is a sequential probability ratio test (SPRT) and the estimation part relies upon a maximum-a-posteriori (MAP) criterion, gated by the detection stage (the parameter to be estimated is the trajectory of the state evolution of the system itself). A thorough analysis of the asymptotic behaviour of the test in this new scenario is given, and sufficient conditions for its asymptotic optimality are stated, i.e. for almost sure minimization of the stopping time and for (first-order) minimization of any moment of its distribution. An application to radar surveillance problems is also examined.