NSA Technology Directorate Analysis of Converged Data

Open PDF
Creating agency: NSA National Security Agency (US)
Document date: 2009
Pages: 1
Genre: slides
Classification: -
Dissemination: -

This document outlines the strategic integration of passive and active signals intelligence (SIGINT) to create a more efficient surveillance loop. The document centers on TURBINE, an intelligent command-and-control system designed to manage a massive network of malware implants (Computer Network Exploitation or CNE). The primary goal is to automate the extraction of data from targeted machines, effectively "converging" traditional passive collection with active hacking to refine intelligence tasking.

Long Description:
This document describes a data flow where information from multiple sources is funneled into centralized repositories for analyst review. Passive collection (unselected metadata and content) and Active collection (forensic data from every implant on a compromised "box") are aggregated into systems like METAWAVE (a data warehouse) and MARINA (a metadata repository). This converged data allows analysts in "SIG Dev" (Signal Development) and "Target Dev" (Target Development) to identify gaps in their knowledge. A critical innovation highlighted is the TURBINE system's "high-level tasking" capability. According to the document, the objective is to decouple the analyst from the technical minutiae of hacking. By automating the "Expert System"—described in the accompanying news article as the "brain"—TURBINE allows a user to request broad categories of information, such as "all details about application X," without needing to understand the underlying file structures, registry entries, or specific data locations on the infected machine. The document confirms that if passive collection is insufficient, the agency will use TURBINE to "pursue active collection." This reflects the "Owning the Net" strategy mentioned in the Intercept article, where the NSA scales from managing hundreds of implants to potentially millions. By using automated "man-on-the-side" attacks (like QUANTUM) to inject these implants, the agency can circumvent encryption and gain "unfettered access" to a device before the data is even scrambled. This automated feedback loop—where passive monitoring tips active implants, which then exfiltrate data back to the NSA.

Analysis of Converged Data

TECHNOLOGY
DIRECTORATE

Converged data leads to better analysis, which leads to more focused tasking

Passive
unselected content
unselected metadata
selected content
TUBE
CNE / Forensics
unselected metadata
(from every implant that is on each box)
METAWAVE (Warehouse)
MARINA (Metadata)
Analyst (SIG Dev / Target Dev)

TURBINE

  • Direct, high-level tasking:
    • Where active tasking already in use, refine what is collected.
    • If no CNE, pursue active collection.

Note: Our goals for high-level tasking are to relieve the user from needing to know / care about the details. For example, a user should be able to ask for "all details about application X" and not need to know how and where the application keeps files, registry entries, user application data, etc.

1