National RF measurement infrastructure · South Africa
South Africa is coordinating spectrum from declarations, not measurements.
The country has committed its dynamic-spectrum-sharing regime to a database-mediated model. That database relies on ICASA-held assignment information and technical information registered by operators and incumbents — records the regulator itself has had to ask to be supplemented. SIGNATA is building the measurement layer underneath it.
The gap
In mid-2024 ICASA issued requests for technical information on systems operating in 3800–4200 MHz and 5925–6425 MHz — and then a further request, seeking information to supplement what the first had received.
The further request shows the regulator required additional incumbent information after the initial round — it does not establish that every existing record was defective, and this site does not claim so. It is, on the public record, the founding observation behind SIGNATA: the records needed measuring against.
Verification status of every regulatory reference on this site is maintained on the regulatory tracker — key instruments are verified against the gazettes themselves; anything not yet verified is marked.
Unregistered incumbents
A missing or inaccurate incumbent record can expose a licensed service to interference — a coordination database cannot protect a user it does not know exists.
Over-declared incumbents
An obsolete or overly conservative record can sterilise spectrum others could use — exclusion zones are calculated from records, not from what is actually transmitting.
No measured evidence base for 6 GHz
The upper 6 GHz band is identified for potential IMT use and carries incumbent fixed-service assignments and, in relevant portions, fixed-satellite allocations. A feasibility study has been committed to. ICASA operates its own fixed spectrum-monitoring stations; we have not found a continuous occupancy dataset published from them, and we have not identified any continuous, independently measured occupancy record for that band available to inform the study.
Registry reconciliation, inspection and enforcement all help with these problems. What none of them provides is evidence of what is actually radiating — that is what measurement adds, and it is the one input the current process does not have.
What SIGNATA is building
A permanent, calibrated, distributed RF sensing network producing continuous occupancy and interference data across 400 MHz – 7.2 GHz — sold as a data service to regulators, incumbents, operators and the spectrum coordination database itself.
| Property | Design target |
|---|---|
| Frequency coverage | 400 MHz – 7.2 GHz, including the full 6.425–7.125 GHz upper 6 GHz band |
| Measurement type | Continuous occupancy statistics, with coherent IQ capture on triggered events |
| Calibration | Traceable chain via transfer-standard nodes |
| Evidentiary handling | Append-only archive, hash-chained, anchored to an external timestamping authority, recomputable by a third party from a verifier and conformance vectors that will be published |
| Retention | Indefinite by design, under scoped and audited access — subject to law, contract and a published retention policy. Historical occupancy cannot be collected retroactively. |
Measured, not modelled
The studies these decisions currently draw on are modelling studies: propagation predictions built on declared transmitter parameters. Modelling is useful, and it inherits every error in the declarations it starts from.
A measurement network starts from what is actually radiating. Where the two disagree, the disagreement is itself the finding — and in a registry the regulator has needed supplementary information requests to complete, disagreement is a case to plan for.
SIGNATA does not operate networks. No WISP, no fixed wireless, no private 5G. The moment a measurement operator sells connectivity, it becomes a competitor to the parties it measures, and the neutrality of the data is gone. That neutrality is the asset.
Where this stands
Pre-formation, in motion. The plan is public, dated, and measured against its own acceptance criteria — the same discipline the product sells.
- Regulatory case — documented and tracked in public on this site, verification status stated per instrument.
- Architecture — two tiers specified, design targets published, every number an acceptance criterion.
- Platform reference implementation — the Phase 0 data pipeline runs end to end: a frozen canonical serialisation with a conformance-vector corpus, an append-only hash-chained evidence store with an offline verifier, the node loop, the ingest boundary and an entitlement-scoped query API. Running on synthetic input while the sensing hardware is built. How it works.
- Architecture decision records — in progress: image strategy, dynamic range and co-site survival, each claim evidence-cited and challenged before any board is committed.
- Prototype sensing node — pending: architecture specified; build and calibrated bench validation to follow.
- Pilot deployment — evidence contracts tied to the upper 6 GHz feasibility study.
The network is not deployed yet — this site says so plainly. If you hold spectrum in 6.425–7.125 GHz, coordinate deployments in 3.8–4.2 GHz, or carry a statutory interference-protection mandate, the conversation is worth having now, while the baseline can still be built — not after the study concludes.