SENSORS & AI
Operators review radar, camera and network data in a command room, illustrating how passive 5G sensing could contribute to an AI-assisted decision chain. Illustrative image; it does not depict a specific NATO exercise or operational system. AiSS Intelligence
NATO’s October experimentation in Latvia offered a useful test of a problem that has become central to counter-drone defence: how to connect sensing, communications, command and response when each component may come from a different nation or supplier. The exercise brought together Digital Backbone eXperimentation (DiBaX26), the Layered Counter-Uncrewed Aircraft Systems Initiative (LCI-X) and Next Generation Targeting (NGT). NATO described the combined chain as one that must sense, connect, understand, decide and act. That formulation matters because a drone detected by a sensor is not yet a threat that a force can manage.
The trials examined resilient communications, counter-UAS integration and the movement of information to decision-makers. NATO’s Communications and Information Agency said DiBaX26 tested 5G network slicing, communications across multiple networks and passive object detection using 5G infrastructure. The culminating FUSE scenario combined live and simulated data, drone swarms, cross-border incursions and simulated electronic warfare. NATO reported participation by more than 30 countries and 100 industry partners, with about 500 personnel across three locations; the culminating demonstration used nearly 400 drones. These figures describe the scale of an experiment, not the performance of an operational network.
The distinction is essential. A demonstration can establish that interfaces can be connected under chosen conditions. It does not, by itself, establish detection probability, track continuity, false-alarm rates, response latency, resistance to deception, or performance when bandwidth, positioning data or spectrum access are disrupted. Nor does the presence of AI agents in an exercise show that an autonomous system can make or execute a legally and operationally acceptable engagement decision. NATO said the results will inform further assessment and capability development; it did not announce that the demonstrated chain has been fielded as an Alliance-wide service.
The most consequential engineering question is therefore not whether one sensor can detect one drone. It is whether a shared track remains usable as it moves across sensors, networks, national command systems and legal jurisdictions. Each hand-off can introduce delay or ambiguity: a sensor may report a track with a confidence score; a fusion system may classify it differently; a command post may lack the context needed to distinguish hostile action from an aircraft in distress; and an effector may be unavailable or prohibited from acting in that airspace. The chain is only as dependable as its weakest transition.
This problem also extends beyond military formations. Ports, airports, energy sites and transport networks may use different detection equipment and have different authorities responsible for safety, security and continuity. A common operating picture can help, but only if data formats, time references, confidence measures and escalation procedures are compatible. Sensor fusion without shared rules can create a larger display of inconsistent information rather than a clearer understanding of risk.
AiSS Assessment
NATO FUSE and DiBaX26 should be read as evidence of a shift in the centre of gravity from individual devices to the reliability of the full sensing-to-action chain. The event’s value lies in exposing integration questions at realistic scale, including spectrum congestion and the use of civilian communications infrastructure. Public information does not yet provide comparable performance data or demonstrate sustained operations under contested conditions. The next test should therefore report not only that systems were connected, but how often tracks persisted across hand-offs, how quickly operators reached a decision, how false tracks were resolved, and what degraded communications did to response quality.
For AI-enabled functions, procurement and doctrine should distinguish assistance from authority. Automated classification and prioritisation can reduce operator workload, but the record should show what data the model used, when confidence fell, how operators could challenge its output and whether the system failed safely. Common test protocols across NATO and European Union programmes would make results more transferable and reduce the risk that a successful demonstration is mistaken for operational readiness. The near-term measure of progress is not the number of sensors or algorithms installed; it is a verified reduction in the time and uncertainty between detection and an accountable response.
