Software Engineering Institute

This poster was presented at FloCon 2023, an annual conference that focuses on applying any and all collected data to defend enterprise networks.

Booz Allen developed Cyber Precog, a GPU-enabled software and data engineering platform that brings operationally honed cyber tooling and a modular pipeline for rapid capability deployment. Cyber Precog supports data ingestion, fusion, and normalization, along with analytic and context enrichment, in one succinct GPU-enabled platform. The Cyber Precog Flyaway Kits serve as the connective tissue between non-kinetic platforms in an operational environment. Additionally, the Flyaway Kits were designed with AI/ML in mind, providing a vital platform for integrating AI/ML models on mission, at the edge.

For operators across the Department of Defense, command and control solutions require both situational awareness and battle management. Precog fills a key gap in the current technical landscape by providing the latter and supporting integration with the former. Many existing solutions fail to aggregate intelligence and data into an actionable manner due to inefficiencies in command and control (C2). Through thoughtfully defined data pipelines and workflow automation, Precog supports mission planning, COA and CONOP development, and streamlined deployment of AI/ML capability.
Joint efforts like the Joint All-Domain Command and Control (JADC2) are centered around connecting sensors, data, and analytic solution into a single solution. The future requires efficiency in sensor-powered awareness and decision-making if commanders are to be more than passive observers of the future battlefield.

Through Precog, AI will help operators to recommend courses of action based on evolving battle conditions and support automation of repeatable and data intensive C2 tasks. AI enables this advancement through:

• Expedited Workflows: Collecting raw data across multiple sensors and platforms to enable AI recommended courses of action based on evolving battle conditions, and automate repeatable, low priority C2 tasks.
• Reduced Cognitive Load: Precog helps reduce the cognitive load on the commander and synthesize battle conditions in real time by supporting AI-enabled decision support that leverages multiple data inputs to support decision making.
• AI Accelerated Data Processing: Precog leverages pre-built mission threads, such as Packet Capture (PCAP) to NetFlow aggregation, enabling forward deployed defenders to efficiently automate their data processing and response. Its 80% GPU acceleration over baseline CPU mission execution allows operators to focus on the hard 20% (APTs, novel malicious attacks, sophisticated anomalous behavior, and well-hidden IOCs) rather than the tediously slow mechanics of transferring all data to the rear.