Validation Budgets

bijux-gnss-receiver owns the runtime validation budgets that compare receiver evidence against known truth. These budgets are not decoration around tests. They are the thresholds that decide whether acquisition, tracking, observation, and PVT evidence is strong enough for receiver-side claims.

flowchart TB
    truth["synthetic or reference truth"]
    run["receiver runtime output"]
    budget["truth_guided_receiver_accuracy_budgets"]
    report["validation report"]
    artifact["navigation accuracy artifact"]

    truth --> report
    run --> report
    budget --> report
    report --> artifact

Budget Families

family	what the budget limits	primary proof
acquisition	Doppler error, Doppler-bin error, code-phase samples, and code-phase chips.	acquisition accuracy-budget test
tracking	Carrier error, Doppler error, code-phase error, and CN0 error over stable epochs.	tracking accuracy-budget test
observation	Pseudorange, carrier phase, Doppler, and CN0 observation errors.	observation accuracy-budget test
receiver clock and PVT	Position error, clock-bias error, residual RMS, and PDOP.	navigation PVT and clock accuracy-budget tests
serialized evidence	Observed maxima and threshold values survive artifact serialization.	navigation accuracy artifact test

Reader Rule

Read the budget report as a comparison between truth, receiver output, and named threshold. A passing report should tell the reader:

• which scenario or truth table was evaluated
• which threshold was applied
• which observed maximum came closest to the threshold
• whether every satellite or epoch passed
• where the same threshold appears in emitted artifacts

If any of those answers are missing, the budget proof is too weak even when the test is green.

Change Rule

• Tightening a threshold requires proof that the receiver consistently remains inside the new bound.
• Loosening a threshold requires an explanation of the physical, simulation, or numerical reason; "the test failed" is not enough.
• Changing a field name, unit, or serialized shape is also a core contract question; inspect Numerical Budgets.
• Changing synthetic truth generation must update the budget proof that consumes it, not only the fixture.

First Proof Check

Start with the receiver reference-validation guide and receiver test guide. Then inspect the synthetic artifact-type source, the synthetic foundation tests, and the accuracy-budget tests for acquisition, tracking, observations, navigation PVT, receiver clock, and serialized navigation evidence.

Boundary Rule

Receiver validation budgets prove runtime behavior. They may consume signal truth, shared core units, and navigation outputs, but they do not redefine those owners. If a budget change depends on altered signal facts, core record meaning, or navigation estimator policy, review that owner in the same change set.