NETAR
Civilian and national level
For airports, ports, power stations, police, border guards and cities. It spots drones, shows operators what is happening, and records every decision.
NETAR & AGÉN connects drone sensors from any manufacturer, combines what they see into one clear picture, and makes sure a named person decides what happens next. Every decision is recorded.
The two authority levels
Same architecture, same fusion, same audit chain, same human gate. What differs is who is authorised and what they may authorise.
Civilian and national level
For airports, ports, power stations, police, border guards and cities. It spots drones, shows operators what is happening, and records every decision.
Military level
For the military. Everything in NETAR, plus the ability for a commander to authorise a response in advance, for a set time, area and type of threat, because some attacks happen faster than anyone can approve them. It connects to military systems and works on the move.
AGÉN is offered to Polish state customers as a development partnership. Supply outside Poland is subject to Polish government authorisation.
Works with sensors from any manufacturer, including radar using the NATO standard, through our own toolkit.
Some sensors can raise an alert; others can only back one up. Early warnings never count as proof.
Each sensor can only speak for itself, so nobody can fake a sensor to raise a false alarm.
Keeps track of drones even when a connection drops or a sensor’s clock drifts.
Handles groups of drones without flooding operators with alerts.
Every alert comes with the conditions at that moment from SIFRA, a note if it looks like a known false signal, and a plain explanation.
What it sees
Ready to check every alert against drone flights approved by Polish air traffic control, as soon as access opens. A match will make it less urgent but never close it, because a drone’s identity can be copied.
Knows where aircraft are, which drones are broadcasting their identity, and where the drone no-fly zones are.
Ready to use ordinary 5G phone masts as a radar, once a network operator provides the data, so it can see drones that send no signal at all.
Ready to use existing radio and TV signals to see drones without transmitting anything itself, so it can’t be found.
Watches above the height normal drone radars cover, for balloons that carry drones and drop them.
In operation
When a drone leaves one site, the next site is warned before it arrives.
The system gives the facts. A person decides, and every decision is signed and recorded for the police.
Real recordings are replayed to test every update and to train operators.
Built to work offline, on tablets and in vehicles, from one site to many. Now in testing ahead of field trials.
The engagement pathway
DESIGNED · REQUIRES STATE AUTHORISATIONAuthority is exercised in advance and in scope, not removed.
Some drone attacks happen in seconds. AGÉN lets a commander decide in advance exactly what may be done, where, and for how long, and switch it off at any moment.
Everything done under that permission is recorded and always shown as automatic, never as a live human decision.
EÒLDÌON is effector-agnostic. AGÉN integrates the effectors a state customer has chosen, through the same governed boundary that controls every other outbound action.
Constraints
A person makes every important decision. Anything done automatically is always labelled as automatic.
Artificial intelligence can help, but it can never raise an alarm, approve anything or hide anything.
The system can say an alert looks false, but it never hides it. The operator always sees everything.
Neighbouring sites can warn each other, but they never share their data or their defences.
It takes observations from sensors made by any manufacturer, weights them by trust and class, fuses them into one picture of the sky, puts SIFRA’s context beside every alert, and holds every consequential action behind an authorised human.
The two authority levels
Same architecture, same fusion, same audit chain, same human gate. What differs is who is authorised and what they may authorise.
Civilian and national level
For critical infrastructure operators, police, border guard, airports and municipalities. Multi-sensor fusion, trust weighting, incident handling and a complete audit chain, with every consequential action held behind a named human decision.
Military level
Everything in NETAR, plus the supervised engagement window: a commander-held, actively confirmed, time-bound authority under which a pre-authorised response may be executed within a stated envelope. AGÉN fits military systems (approved incidents go to TAK, radars connect over ASTERIX and sensors over SAPIENT) and it works on the move, with convoys and mobile units.
AGÉN is offered to Polish state customers as a development partnership. Supply outside Poland is subject to Polish government authorisation.
Radar, RF, EO/IR, acoustics, Remote ID and ADS-B in one picture, and ready to integrate with fibre, 5G networks, passive radar and citizens. Sensors connect through our own SDK and edge agent, over ASTERIX, or natively over SAPIENT (BSI Flex 335 v2.0), the UK and NATO sensor-integration standard. A customer who already owns a SAPIENT sensor, such as an AESA radar, connects it without a custom integration.
First-class sensors observe the object itself (its airframe, echo, heat or broadcast identity) and can anchor an incident. Second-class sensors, such as acoustics and fibre, only corroborate. Early-warning cues point the way but never count towards confirmation. A fast alert always needs at least one first-class anchor.
Each sensor holds its own credential and can speak only for itself. A phone cannot pose as a radar. Sensors sharing a mast are recognised as one installation, so a single point of failure never counts as independent corroboration.
Drifting sensor clocks and dropped links do not break tracks. Reports held during an outage rejoin the track they describe when the link returns, and each sensor’s clock offset is measured and shown to the operator.
Coordinated groups are tracked as groups, so the operator’s queue doesn’t become the weak point. A group can never raise an incident its members could not raise individually.
Each sensor is read once: the reading goes into SIFRA’s record, and the same observation reaches NETAR under the sensor’s own identity. Every incident carries the conditions at that moment, GHOST HUNTER’s view of whether it resembles a known false signal, and URIEL’s explanation of what is happening around the site. Context explains an incident; it never creates, scores or suppresses one.
What it sees
Built to check every incident against flights authorised through PANSA’s PansaUTM as soon as PANSA opens access. A match will lower priority but never close an alert, because a drone’s identity can be copied. When the feed is unavailable, the record says the system could not check.
The air traffic picture over ASTERIX, ADS-B, Remote ID and UAS geographical zones (ED-269, ED-318). Crewed aircraft are known, permitted drones are recognised, and a zone raises priority, never the threat score.
Ready to integrate with 5G networks used as radar. Base stations already light up the sky; once a network operator provides the echoes, NETAR uses them to see drones, including those that transmit nothing, with no new transmitters.
Ready to integrate with passive coherent location (PCL), which uses existing radio and TV transmitters as its illumination. The receiver emits nothing itself, so it cannot be located, and it sees drones that stay radio-silent.
Balloons from Belarus have closed airspace over north-eastern Poland and shut Vilnius airport. The carrier drifts above the band ground C-UAS radars are set up to watch. NETAR tracks it, recognises weather sondes by their telemetry and recommends higher readiness at release, and the dropped drone enters normal detection.
In operation
When a drone leaves an area, the next site is warned with its course, speed and possible arrival time. An early cue, never confirmation, and never a shared database. The commander view spans many sites, and mobile units receive the same warning on the move.
Counts, not verdicts: “unknown” never poses as known. Every decision has an author and goes into a signed log, which produces the report for police and prosecutors.
SIFRA’s recordings of real conditions are replayed to test every NETAR update before it is released, and to train operators on real events rather than invented ones.
Offline maps, rugged tablets and mobile units. Flight-path prediction, time to asset, line of sight and coverage. Manipulation detection, false-alarm analytics, a citizen app, TAK/ATAK output and a commander view across sites. In testing ahead of field trials.
The engagement pathway
DESIGNED · REQUIRES STATE AUTHORISATIONAuthority is exercised in advance and in scope, not removed.
Modern incursions can close faster than an approval chain. AGÉN is built for that. A commander defines the envelope — what may be engaged, under what conditions, within what geography, for how long — confirms it actively, and can revoke it at any moment. Within that envelope, and subject to authorisation, AGÉN will be able to execute at machine speed.
Every action is recorded, attributed, and permanently distinguishable from an operator-approved one. Nothing executed under a standing authority can ever be presented as a live human decision.
The structure this requires already exists: the gate conditions, the commander’s signed authority, the audit chain and the governed outbound boundary. Building the engagement behaviour itself is a funded, counsel-led, state-partnered decision, validated to a stricter standard than civilian deployment requires.
EÒLDÌON is effector-agnostic. AGÉN integrates the effectors a state customer has chosen, through the same governed boundary that controls every other outbound action.
Constraints
No external authority is notified and no effector is cued without a live named-human decision or a commander-signed standing order. An automated action is permanently distinguishable in the record from an approved one.
The rules-based scorer is authoritative. A machine-learning classifier may contribute one bounded, recorded, default-off factor to a threat score. It cannot create an incident, approve one, dispatch anything or suppress anything.
Artefact assessment ranks and explains. It never removes a detection from an operator’s view.
Peer notification between sites is one-way, and sites share no database. The payload describes the object and never the observer. A relayed observation can never corroborate.