Copula based dependence modeling for inference in RADAR systems

Abstract

Statistical dependence is one of the significant design issues in various radar systems for inference tasks including detecting an activity of interest or estimating states or parameters for situational awareness. Modeling dependence has been discussed in many articles on radar and the research has shown that taking dependence into account improves performance of inference tasks. In this paper, we introduce copulas as flexible tools for modeling of nonlinear/linear dependence. Copulas allow one to model the dependence structures among random variables with arbitrary marginal distributions. We explore the potential use of copula theory in radar systems while discussing the dependence modeling problem. Then we present an application for binary hypothesis testing to show the benefit of using copula theory.