Mixture Modeling based Probabilistic Situation Awareness

TL;DR

This paper introduces probabilistic models, including HMM and ESSM, utilizing Gaussian mixture models to improve real-time situational awareness by modeling the evolution and observable relationships of situations.

Contribution

It presents a novel application of mixture modeling within HMM and ESSM frameworks for dynamic and real-time situational awareness.

Findings

ESSM outperforms HMM in real-time estimation

Models effectively capture the evolution of situations

Simulation confirms the approach's efficiency

Abstract

The problem of situational awareness (SAW) is investigated from the probabilistic modeling point of view. Taking the situation as a hidden variable, we introduce a hidden Markov model (HMM) and an extended state space model (ESSM) to mathematically express the dynamic evolution law of the situation and the relationships between the situation and the observable quantities. We use the Gaussian mixture model (GMM) to formulate expert knowledge, which is needed in building the HMM and ESSM. We show that the ESSM model is preferable as compared with HMM, since using ESSM, we can also get a real time estimate of the pivot variable that connects the situation with the observable quantities. The effectiveness and efficiency of both models are tested through a simulated experiment about threat surveillance.