axonos-signal-pipeline v0.6.0 — calibration layer

axonos-signal-pipeline v0.6.0 — calibration layer

Completes a three-layer release that ships together:

• 0.4.0 deterministic fixed-point features (variance, log-variance, RMS, abs-mean, zero-crossings; isqrt / log2_q16 primitives)
• 0.5.0 classifier inference (minimum-distance-to-mean, linear/LDA with confidence + abstain) — caller-supplied parameters, no trained model
• 0.6.0 calibration (channel covariance, session mean, drift update, Cholesky reference whitening W·R·Wᵀ = I, ZeroCalib skeleton)

Every stage is deterministic machinery pinned by conformance vectors (vectors/pipeline-vectors-v0.6.0.json). #![no_std], allocation-free, #![forbid(unsafe_code)], zero dependencies.

No trained model, no measured accuracy / latency / power, no clinical claim. Engineering demonstrator — not a medical device.

AxonOS Signal Pipeline v0.2.4 — Integer DSP Primitives

AxonOS Signal Pipeline v0.2.4 — Integer DSP Primitives

Public release for the AxonOS reference signal-pipeline contract.

Highlights

• Adds deterministic integer DSP primitives:
  • DC / mean removal.
  • Fixed-point causal FIR engine.
• Adds typed DSP errors:
  • EmptyInput
  • EmptyKernel
  • OutputLengthMismatch
  • InvalidShift
• Updates conformance vector set to 0.2.4.
• Adds dc_remove and fir conformance cases.
• Updates generated Rust test data.
• Fixes CI-blocking Clippy type-complexity in generated vector cases.
• Keeps the project in pre-clinical engineering scope.

Verification

Latest CI is green.

Validated gates include:

• cargo fmt --all --check
• cargo clippy --workspace --all-targets -- -D warnings
• cargo test --workspace --locked --all-targets
• cargo build -p axonos-pipeline-core --target thumbv7em-none-eabihf
• python3 tools/validate_vectors.py
• python3 tools/check_hygiene.py

What This Release Adds

remove_mean:

• deterministic integer DC removal,
• mean truncated toward zero,
• saturating subtraction.

fir:

• deterministic fixed-point causal FIR engine,
• i64 accumulator,
• arithmetic right shift,
• round-half-up bias,
• i32 saturation.

Not Claimed

This release is not:

• a medical device
• clinical software
• a validated band-pass or notch filter
• a classifier accuracy claim
• hardware BCI integration
• patient-data processing
• diagnosis or treatment software

fir is a deterministic fixed-point convolution engine. It does not claim any validated frequency response.

axonos-signal-pipeline v0.1.0

Initial release: the type contract and conformance surface for the AxonOS
reference signal pipeline.

Added

• axonos-pipeline-core crate (#![no_std], #![forbid(unsafe_code)],
  zero dependencies) with the typed stage contract:
  • RawFrame — validated raw acquisition frame, time-major interleaved,
    column-compacted 24-bit samples; FNV-1a 64 integrity checksum; Debug
    redacts sample values.
  • ChannelMask, SampleRate newtypes.
  • Epoch / EpochIter deterministic windowing with ExactSizeIterator.
  • artifact_scan amplitude/saturation screening (pure integer).
  • FeatureVector<D> placeholder type.
  • ClassifierDecision pipeline-terminal type.
  • Sealed BoundarySafe trait — only ClassifierDecision may cross the
    application boundary; raw types rejected at compile time.
• Conformance vectors vectors/pipeline-vectors-v0.1.0.json and synthetic
  fixture fixtures/synthetic/frame-0001.json, generated by
  tools/gen_test_vectors.py and integrity-pinned by vectors/SHA256SUMS.
• CI gates tools/validate_vectors.py (exact reproducibility) and
  tools/check_hygiene.py (contact-metadata hygiene).
• Documentation: pipeline contract, claims and evidence levels, limitations,
  privacy boundary, validation plan, calibration design note.

Notes

• This release implements no DSP, feature extraction, classifier, or
  calibration; those are typed placeholders introduced behind conformance
  vectors on the roadmap (v0.2.0–v0.5.0). No accuracy, latency, or power
  figure is claimed. Pre-clinical engineering artifact; not a medical device.

---

The AxonOS Project · axonos.org · connect@axonos.org · security@axonos.org · github.com/AxonOS-org