PA-KDNC: taking Gridnorth messages over the air

The problem

Gridnorth's results so far stop at the cable. The laboratory evaluation measured message structure, field accuracy and encoded size, and the DDIL page states plainly that nothing has yet been transmitted over a radio link. Airtime figures are arithmetic until they are measured.

Real links behave differently from a bench. Signals fade and drop. Interference arrives without warning. A message that decodes perfectly on a cable may need several attempts over the air, and every attempt adds time on air. For a unit in a contested area, time on air is exposure: the longer a transmitter is active, the longer it can be detected and located.

The approach

1. 01

   Two stations

   Each has a software-defined radio, an antenna suited to the band, filtering and, for longer ranges, a small licensed amplifier.

2. 02

   The same pipeline as the laboratory

   An operator enters a report by voice or text. Gridnorth produces a validated message, encodes it with schema-informed EXI, and hands it to the radio.

3. 03

   Standard modulation

   Messages are carried on established, documented digital modulations. PA-KDNC does not invent a new waveform. The question is how validated, compact messages behave on real links, not how to build a radio.

4. 04

   Stepped testing

   Cable first, then short range, then longer ranges, different terrain and times of day, and deliberately degraded conditions such as reduced power and added noise.

5. 05

   Everything logged

   Bytes sent, time on air, retries, delivery, end-to-end latency and whether each received message is identical in content to the one sent.

Design constraints

Every gain in this problem is paid for somewhere else. These are the tensions the programme has to answer.

Fidelity against airtime. Extraction decides what not to transmit. A report reduced to its essential fields has lost something, so what was extracted must stay visible to the operator who owns the report. Gridnorth returns uncertain fields to the operator before release for this reason.

Inference against power. A model small enough to run on carried equipment shares a battery and a thermal envelope with the radio it feeds. Model size is bounded by what can be carried and cooled, not by accuracy alone.

Compaction against conformance. The most compact encoding of a message is always a private one. Conforming to a published catalogue costs bits a proprietary scheme would not spend. The cost is worth paying, because a report has value only where it can be read. PA-KDNC uses schema-informed EXI, a W3C standard, so any receiver with the published schema can decode the message.

Delivery against exposure. Retransmission improves delivery and increases time on air. How many attempts to make, and when, is a question of survivability before it is a question of networking.

Built against published standards

Reporting leaves the operator in formats the alliance already reads.

• ADatP-3 and APP-11: the NATO message text formatting rules and message catalogue. Gridnorth's structured output targets this catalogue rather than a private schema. See Standards.
• EXI: the W3C Efficient XML Interchange format, used to shrink messages for narrow links.
• Cursor on Target and the TAK ecosystem: the event format and the clients (ATAK, WinTAK, iTAK) where many dismounted operators read the picture. Integration is planned, not built.

Licensing

Over-the-air trials will be carried out under an ISED developmental licence, which provides time-limited access to spectrum for research, experimentation and demonstration. The licence has not yet been requested. Trials will not use military frequencies or equipment.

How it connects to Gridnorth

PA-KDNC is the first implementation of the Gridnorth framework: the platform and radio integration around Gridnorth's extraction, verification and encoding. What PA-KDNC measures on the air feeds back into how Gridnorth prioritizes, retries and sizes messages.

What we will measure

• Delivery rate by range, band and condition
• End-to-end latency, from end of input to message delivered
• Time on air per message, including retries
• Whether each received message is identical in content to the one sent
• Operator workflow outdoors, including time to confirm flagged fields

What we have not done

• Transmitted over any radio link. All published figures are laboratory figures.
• Requested or received an ISED developmental licence.
• Tested on military frequencies or equipment, or integrated with TAK clients.
• Measured any reduction in detectability. PA-KDNC makes no claim about low probability of intercept or detection.

Next milestone

Cable-connected two-station tests, followed by a developmental licence application in the fourth quarter of 2026.

Collaborate

We want to hear from research partners, radio and C2 integrators, and organizations that could host outdoor trials. Contact us.

Last reviewed September 29, 2026