Athena: A Multi-INT Data-Driven Indications and Warning (I&W) Tool for Event Recognition in SSA

Analysts and operators need to know as soon as possible when their satellites are in danger, so that responses can be initiated to provide tactical protection. Numerica proposes to leverage its advanced prototype software, Athena, to provide timely and reliable indications and warnings (I&Ws) derived from evolving situations by discovering anomalous correlations across events detected using multi-source, multi-INT data. The Athena software prototype currently uses state-of-the-art data-driven unsupervised and semi-supervised machine learning methods for joint normality modeling and anomalous event recognition. It was developed under previous AFRL funding, and validated using real photometric data collected from electro-optical sensors (i.e., light-curve data, an example of measurement and signature intelligence, or MASINT), as well as certain open source structured datasets (structured OSINT). The core technology has also been applied by Numerica to detect distributed anomalies and advanced persistent threats (APTs) in cyber networks. Under the proposed effort, Numerica would upgrade Athena to exploit additional types of multi-INT data (including SIGINT), with the goal of demonstrating reduced false alarm rates and increased probability of anomalous event recognition. The Phase I program will culminate with a proof-of-concept demonstration of the I&W tool using combinations of real-world and simulated datasets.