Prediction estimates performance. Exact computation enforces constraints.

The engine separates physics, prediction, optimisation and explanation. The k-NN uses a versioned scientific dataset to compare eight LoRa and LR-FHSS modes. Results remain simulation-based estimates, with no real-link validation to date.

Design principle

Predictors estimate performance. The deterministic engine decides under constraints.

State & mission

TLE or manual state, radio scenario, active-node count, priority and policy.

Prediction

PER and available metrics by mode. Confidence is shown only when provided and qualified.

Exact operator

Filters constraints and chooses the optimal admissible option.

Explanation and analysis

Presents factors and warnings without changing the decision. Agent coordination remains future.

8radio modes compared

Exactconstraints checked before selection

Versionedidentified scientific dataset

Auditabledecision trace and limitations

V1 benchmarks provide a scientific foundation. Results depend on the dataset and protocol; they do not guarantee field performance.

From V1 to product

The next step is validation with real data, modems and networks.

Propose a partnership