Intelligence Fusion FAQ
Fast answers to the questions people ask about intelligence fusion. For the full treatment, see What Is Intelligence Fusion?.
What is intelligence fusion?
Intelligence fusion is the practice of combining many sources into one coherent, decision-ready picture. It turns fragmented, overlapping, sometimes contradictory inputs into an assessed understanding of what is out there, what it means, and what to do about it. It is the broadest term in a family that also includes data fusion, sensor fusion, and information fusion.
What does intelligence fusion mean in plain terms?
It means taking everything you can see about a situation, from many different kinds of source, and producing one clear answer instead of many noisy ones. A radar track says something is there. Fused with ownership, environment, and pattern of life, it tells you what it is, whose it is, and whether to care. That shift from data to understanding is the whole point.
What is the difference between data fusion, sensor fusion, information fusion, and intelligence fusion?
Data fusion combines low-level data such as signals, pixels, and detections. Sensor fusion combines the outputs of multiple sensors into tracks and identities. Information fusion combines processed information across sources. Intelligence fusion is the broadest of the four and combines all of it, across disciplines and domains, into assessed intelligence an operator can act on.
What is the JDL data fusion model?
The JDL (Joint Directors of Laboratories) model is the standard framework for describing fusion as a stack of levels from raw signal to decision. The widely used revision, extended by the DFIG (Data Fusion Information Group), defines six levels, 0 through 5: signal assessment, object assessment, situation assessment, impact assessment, process refinement, and user refinement. It is the common vocabulary the whole field uses, even when practitioners do not name it.
What are the levels of the JDL/DFIG model?
Level 0 is signal and sub-object assessment, the raw conditioning of detections. Level 1 is object assessment, turning detections into tracks and identities. Level 2 is situation assessment, understanding how objects relate. Level 3 is impact assessment, what it means and what the threat or opportunity is. Level 4 is process refinement, the system tuning its own collection and processing. Level 5 is user refinement, the human and cognitive loop that keeps the picture aligned to what the operator actually needs.
What is multi-INT fusion?
Multi-INT fusion combines multiple intelligence disciplines, the INTs, into one assessed picture: geospatial, signals, imagery, measurement and signature, open-source, human, and financial intelligence among them. It is distinct from multi-sensor fusion, which combines multiple sensors of the same general kind into tracks. Multi-INT is about crossing disciplines, not just modalities.
What is the difference between multi-INT and all-source intelligence?
The terms overlap heavily. All-source is the traditional intelligence-community phrase for analysis that draws on every available discipline to produce a finished assessment. Multi-INT is the more engineering-flavored term for fusing those disciplines, often in software, often in or near real time. In practice multi-INT is how all-source gets built at machine speed.
What is the difference between sensor-to-shooter and sensor-to-sensor fusion?
Sensor-to-shooter is getting a detection to an effector fast enough to act, the compressed kill chain. It is real but it is one path through the stack. Sensor-to-sensor fusion uses one sensor to cue another, cross-correlating modalities so the picture improves even when nobody is shooting: a passive RF hit cues a radar, an acoustic detection cues an optic, a financial flag cues a closer look at a vessel. Most fusion is about building understanding, not firing solutions.
Is data fusion only a military thing?
No. The fusion problem is universal, and most of the people who have it have never used the word JADC2. A wildfire incident command post fusing aircraft positions, crew locations, weather, and terrain is doing all-domain fusion. A search-and-rescue coordinator correlating AIS, drift models, and last-known-position is doing situation and impact assessment. A port authority watching vessels, manifests, and sanctions exposure is doing multi-INT fusion.
What is data fusion in a military context?
In the military it is the substance of command-and-control modernization. JADC2 (Joint All-Domain Command and Control) is the Department of Defense approach to connecting sensors, deciders, and effectors across land, maritime, air, space, and cyberspace into one decision enterprise, often summarized as Sense, Make Sense, Act. Fusion is the "Make Sense" layer that turns cross-domain collection into understanding faster than an adversary can act.
What makes intelligence fusion hard?
Sources arrive in different formats, at different latencies, with different confidence models and different classifications. The same real-world entity shows up under different identifiers in different feeds, so resolving identity is a problem in its own right. And contradiction is normal: two sources will disagree, and the system has to resolve which to believe and carry the provenance so a human can audit the call. Drawing a dot on a map is easy; producing a trustworthy, explainable track is the hard part.
Where does Empyrean fit?
Empyrean fuses across sensors, identity and finance, environment, narrative, and space, through all six levels of the recognized data-fusion model, unclassified and deployable to the edge. The output is a fused track with identity provenance, not an uncorrelated map. Details on the Fusion capability page.
What is a multi-INT fusion platform?
A multi-INT fusion platform is software that ingests, correlates, and fuses data from multiple intelligence disciplines (GEOINT, SIGINT, IMINT, MASINT, OSINT, HUMINT, FININT) into a unified operational picture. Unlike single-discipline tools that excel within one INT, a multi-INT platform performs cross-domain entity resolution, maintains identity provenance across sources, and produces assessed tracks rather than uncorrelated layers on a map.
The defining characteristic is mathematical correlation across disciplines, not just visual co-display. A map that shows a radar track and a financial alert side by side is not fusion. A platform that resolves those two observations into the same entity, with a scored confidence and an auditable provenance chain, is.
Key capabilities of a multi-INT fusion platform include:
- Cross-domain entity resolution: Determining that a vessel on radar, a beneficial owner in a corporate registry, a SIGINT intercept, and an OSINT press release all refer to the same real-world entity.
- Identity provenance: Tracking how identity was established, which sensors contributed, whether identity was claimed, observed, or inferred, and how confidence has evolved.
- Temporal correlation: Connecting observations separated in time into a coherent behavioral pattern (pattern of life, pattern of trade, pattern of communications).
- Multi-classification handling: Operating across classification boundaries where some feeds are unclassified, some are FOUO, and some are classified, without contaminating lower-classification outputs.
- Edge deployment: Operating in DDIL environments at the tactical edge without persistent cloud connectivity.
Empyrean is a multi-INT fusion platform. It fuses across sensors (radar, RF, EO/IR, AIS, ADS-B), identity and finance (entity resolution, beneficial ownership, sanctions), environment (weather, terrain, propagation), narrative (social media, OSINT, information operations), and space (orbital tracking, conjunction assessment). See Fusion.
What are alternatives to multi-INT fusion platforms?
Alternatives to multi-INT fusion platforms span a spectrum from fully manual analysis to partial automation, each with trade-offs against integration quality and operational tempo.
- Single-discipline tools: One tool per INT (a maritime AIS tracker, a SIGINT workstation, an OSINT aggregator). Each excels within its domain but produces siloed outputs that the analyst must manually correlate. This is the dominant model in most organizations and is the status quo that multi-INT platforms replace.
- Analyst-driven manual collation: Analysts build link charts, timelines, and assessments by hand using general-purpose tools (Palantir Gotham, Analyst's Notebook, spreadsheets, wikis). Quality depends entirely on analyst skill and available time. Does not scale to real-time operations or high-volume data.
- Federated search: Query interfaces that search across multiple databases without moving or fusing the underlying data. Useful for discovery but does not perform entity resolution, temporal correlation, or confidence scoring.
- Dashboard aggregators: Platforms that display multiple feeds on a common screen without mathematical fusion. Tracks from different sources appear as separate objects. The operator performs the mental fusion. Works for low-density environments but breaks down when track density exceeds human cognitive capacity.
- Custom integration via APIs: Engineering teams build bespoke pipelines connecting various tools. Can achieve high-quality fusion but requires significant development investment, ongoing maintenance, and rarely provides a general-purpose fusion engine.
The core trade-off: simpler alternatives reduce cost and complexity but leave the analyst as the fusion engine. As threat density, data volume, and operational tempo increase, the human bottleneck becomes the constraint. Multi-INT platforms move the computational burden from the operator to the machine while preserving human oversight at the decision layer.
What is data fusion in intelligence analysis?
In intelligence analysis, data fusion is the automated or semi-automated combination of raw and processed data from multiple collection disciplines into correlated entities, relationships, and assessments. It spans the JDL/DFIG model from signal-level processing (Level 0) through situation assessment (Level 2) and impact assessment (Level 3).
The practical output is the difference between an analyst receiving 47 separate reports about maritime activity in the Taiwan Strait and receiving a single assessed picture showing 12 resolved entities with identity, ownership, behavior classification, and anomaly flags. The analyst still makes the judgment call, but they are judging a picture rather than assembling one from fragments.
Data fusion in intelligence analysis typically involves:
- Entity resolution: Determining which records across multiple databases refer to the same real-world entity.
- Track correlation: Associating detections from multiple sensors into a single kinematic track with identity.
- Activity pattern recognition: Identifying behaviors (loitering, transshipment, route deviation, pattern-of-life change) that match intelligence indicators.
- Confidence scoring: Assigning and maintaining a quantitative assessment of how much to trust each fused conclusion, based on source reliability, corroboration, and recency.
- Provenance tracking: Maintaining the chain of evidence from raw collection through every processing step to the final assessment, so the analyst can audit any conclusion.
For Empyrean's approach to intelligence fusion, see What Is Intelligence Fusion? and the Fusion capability page.
