Files
ruvnet--RuView/v2/crates/wifi-densepose-calibration/tests/full_loop.rs
T
rUv 17471e93ff ADR-152: WiFi-Pose SOTA 2026 intake — WiFlow-STD benchmark, Rust integrations, ADR-153 802.11bf layer, efficiency frontier (#1008)
* feat(calibration): NodeGeometry transceiver-geometry recording (ADR-152 §2.1.1)

PerceptAlign-motivated geometry capture at enrollment: per-node optional
records (position, antenna orientation, inter-node distances, acquisition
method) — recorded when known, never required. Event-sourced via
EnrollmentEvent::GeometryRecorded (latest recording wins); persisted on
SpecialistBank with serde defaults so pre-ADR-152 bank JSON loads cleanly
(fixture-proven, and geometry-free banks serialize byte-shape-identical
to the old schema); threaded through MultiNodeMixture as data only — the
learned geometry embeddings and algorithmic fusion use are §2.1.2,
deliberately deferred until the ADR-151 P6 LoRA heads exist.

Geometry recorded from now on means banks captured today remain usable
for layout-conditioned training later — you can't retroactively add
geometry to data you didn't record.

8 new tests (3 geometry, 2 anchor, 2 bank, 1 multistatic) + full-loop
extension (2-node geometry, one tape-measured + one unknown, surviving
the bank JSON round-trip the runtime loads from). 50/50 calibration
(both feature configs) + 23 CLI tests green.

Co-Authored-By: RuFlo <ruv@ruv.net>

* feat(training): two-checkerboard camera↔room calibration for ADR-079 labels (ADR-152 §2.1.3)

Defends the camera-supervised pipeline against PerceptAlign's
"coordinate overfitting": MediaPipe keypoints were emitted in raw camera
coordinates with no shared frame and no transceiver-geometry metadata —
the exact label shape that memorizes deployment layout and collapses
cross-layout.

- scripts/calibrate-camera-room.py + calibration_lib.py: OpenCV
  two-checkerboard calibration → versioned bundle JSON (intrinsics,
  camera→room extrinsics, checkerboard spec, transceiver geometry,
  sha256 calibration_id). Intrinsics resolve from file > cache >
  multi-view computation > loud-warning 2-view fallback.
- collect-ground-truth.py --calibration <bundle>: every sample gains
  keypoints_room (unit bearing rays from the camera center in the room
  frame — documented projective alignment; raw image coords preserved
  so training chooses), camera_origin_room, calibration_id, and the
  transceiver geometry stamp. Without the flag, output is byte-identical
  to before (tested) + a one-line ADR-152 warning.

Design finding (recorded for ADR-152): a single planar checkerboard's
corner grid is centrosymmetric — the reversed corner ordering fits a
ghost camera pose with IDENTICAL reprojection error, so per-board flip
disambiguation is mathematically ill-posed. solve_two_board_extrinsics
solves the joint wall+floor set over all 4 flip combinations, where the
minimum is unique — an independent reason the TWO-checkerboard method is
required, beyond what PerceptAlign states.

15 headless pytest tests green (synthetic corners: extrinsics recovery
incl. ghost resolution, bundle round-trip + hash stability, ray
transforms w/ distortion + cross-resolution, no-calibration byte
identity).

Co-Authored-By: RuFlo <ruv@ruv.net>

* feat(benchmarks): WiFlow-STD reproduction harness + measurement (a) results (ADR-152 §2.2)

Shipped checkpoint REFUTED (0.08% PCK@20, wrong keypoint normalization);
6 reproducibility defects documented (broken imports, corrupted dataset
tail with float32-max garbage that NaN-poisons fp16 BatchNorm, unreachable
test phase). After repairs, retraining with upstream defaults reproduces
96.09% PCK@20 full-test / 96.61% corruption-free (published 97.25%) on
RTX 5080. Claims graded MEASURED-EQUIVALENT; 2.23M params + ~0.055 GFLOPs
verified. Third-party code/weights/data stay out of tree (gitignored).

Co-Authored-By: claude-flow <ruv@ruv.net>

* feat: ADR-152 Rust integrations + ADR-153 802.11bf protocol model

- calibration: GeometryEmbedding — 32-slot permutation-invariant NodeGeometry
  featurization for future LoRA-head conditioning (ADR-152 §2.1.2); derived
  SpecialistBank::geometry_embedding() accessor; 59 tests
- train: MaePretrainConfig + patchify/random-mask with UNSW measured recipe
  (80% masking, (30,3) patches; ADR-152 §2.3, arXiv 2511.18792); strict
  no-truncate/no-NaN policy; proptest properties
- train: WiFlowStdModel — tch-gated port of the verified ~96%-PCK@20
  WiFlow-STD architecture (ADR-152 §2.2 beyond-SOTA); ungated param formula
  pinned to 2,225,042; 15/17-keypoint support; 239 crate tests
- hardware: ieee80211bf forward-compatibility protocol model (ADR-153):
  SpecProfile gates, SensingCapabilities negotiation, required ConsentMode,
  session FSM, SensingTransport + SimTransport + OpportunisticCsiBridge;
  full acceptance checklist covered; 156+4 tests
- deps: ruvector bumps per ADR-152 §2.6 survey (mincut/solver 2.0.6,
  attention 2.1.0, gnn 2.2.0); vendor/ruvector synced to a083bd77f
- docs: ADR-153 accepted; ADR-152 §2.2 status, §2.4 amendment, §2.6 added

Workspace: 162 test suites green (--no-default-features); Python proof PASS.
Known pre-existing flake: homecore-api env_empty_falls_back_to_defaults
(unserialized env-var mutation) — untouched, follow-up.

Co-Authored-By: claude-flow <ruv@ruv.net>

* docs: CHANGELOG + CLAUDE.md entries for ADR-152 integrations and ADR-153

Co-Authored-By: claude-flow <ruv@ruv.net>

* fix(train): repair tch-backend bit-rot — gated path compiles and tests run again

Mechanical API refresh against current tch: Vec::from(Tensor) -> try_from
(+ explicit flatten), numel() usize cast, Rem/div ops -> remainder() /
divide_scalar_mode(floor) — the latter fixed a silent true-division bug in
heatmap argmax decoding; clamp(1.0, f64::MAX) -> clamp_min (torch 2.x scalar
overflow panic); petgraph EdgeRef import; missing EvalMetrics and
verify_checkpoint_dir APIs that tests documented. wiflow_std roundtrip test
uses safetensors (.pt _save_parameters roundtrip broken in torch 2.11
Windows). Gated: 349 passed (incl. all 20 wiflow_std); ungated: unchanged.
Known pre-existing: gaussian-heatmap convention mismatch (2 tests), proof
seed race under parallel threads — documented, deliberate follow-ups.

Co-Authored-By: claude-flow <ruv@ruv.net>

* feat(train): WiFlow-STD PyTorch->tch weight import + numerical parity proof

export_to_safetensors.py maps the retrained checkpoint (295 tensors -> 248
mapped, param sum exactly 2,225,042; num_batches_tracked dropped) into a
tch-loadable safetensors plus a deterministic parity fixture. Gated #[ignore]
integration test loads it strictly and asserts forward-pass agreement:
max abs diff 1.192e-7 on the seed-42 fixture. dump_variable_names test makes
the tch name layout authoritative. Zero architecture discrepancies found.

Co-Authored-By: claude-flow <ruv@ruv.net>

* fix: workflow-review findings — BN gamma init, ThresholdParams serde, init docs

Concurrent validation workflow (2 review lanes + adversarial verification,
13 agents): 5 confirmed findings, 3 refuted. Fixes:
- wiflow_std: pin BatchNorm gamma to 1.0 (tch default draws Uniform(0,1) —
  silently halves activations in from-scratch training; loaded checkpoints
  unaffected, parity re-verified after the change)
- wiflow_std: document the conv-init divergences vs the reference's
  effective kaiming_normal(fan_out) re-init (from-scratch dynamics only)
- ieee80211bf: ThresholdParams deserialization validates via try_from so
  the <=100 invariant holds for untrusted payloads (+ rejection test)

Benchmarks (release, ruvzen): GeometryEmbedding 1.84us/call (542k/s),
MAE tokenization 7.38us/window (135k/s), 802.11bf FSM 8.9M events/s —
nothing suspicious.

Co-Authored-By: claude-flow <ruv@ruv.net>

* docs(adr): ADR-152 §2.1.4 gate resolved — PerceptAlign repo MIT, dataset on HF

Co-Authored-By: claude-flow <ruv@ruv.net>

* feat(benchmarks): edge optimization measured + measurement (b) blocked + 92.9% retraction

Edge optimization (ADR-152 optimize track): ONNX Runtime fp32 is the CPU
latency win (3.2 ms/window, ~3.4x faster than torch, parity 2.4e-7); ORT
dynamic int8 reaches 2.44 MB (paper's ~2.2 MB claim plausible only via
conv-capable toolchains; -0.16pt PCK@20, +18% MPJPE, 2x slower); torch
dynamic quant converts 0% of this conv-only model; fp16 halves storage free
but is slower on CPU.

Measurement (b) BLOCKED-ON-DATA: only 1,077 paired ESP32 windows exist
(stop rule <2k). Forensic recheck of the surviving April holdout RETRACTS
the ADR-079 '92.9% PCK@20' figure: constant-output model, absolute (not
torso) threshold, 69 near-static frames — mean predictor scores 100% under
that protocol; torso-PCK@20 is 19.1%. Corroborates PR #535. Stale citations
removed from user-guide, readme-details, ADR-152 §2.1.3; no-citation rule
extended to ADR-079 accuracy claims. Unblock: >=2k-window multi-pose paired
session + torso-PCK re-baseline.

Co-Authored-By: claude-flow <ruv@ruv.net>

* docs(user-guide): corrected camera-supervised collection tutorial

Step 0 CSI-rate check + session-length math (window yield = frames/20 —
the May session's 8x under-delivery was a ~12 Hz CSI rate, not an aligner
bug); two-checkerboard calibration step (ADR-152 §2.1.3); pose-variety and
confidence guidance; torso-normalized PCK + temporal-split + pred-variance
eval protocol (lessons from the 92.9% retraction); scale presets re-keyed
to realistic window counts.

Co-Authored-By: claude-flow <ruv@ruv.net>

* feat(benchmarks): static PTQ int8 (calibrated) results + overnight capture script

Conv-only static QDQ beats dynamic int8 on accuracy (PCK@20 96.61-96.63%
vs 96.52%, MPJPE +10% vs +18% over fp32) at ~equal size/latency; all-ops
QDQ strictly worse (int8 activations through attention glue). Entropy
calibration verified bit-identical to MinMax on this data. Deployment:
ONNX fp32 for speed (3.2ms), static conv-only QDQ for smallest (2.53MB).

Also: scripts/overnight-empty-capture.py — segmented UDP CSI recorder for
empty-room baselines (no glob collisions, detach-safe).

Co-Authored-By: claude-flow <ruv@ruv.net>

* feat(benchmarks): measurement (b) MEASURED — optimization transfer only, mean-pose baseline wins

WiFlow-STD fine-tuned on 2,046 fresh single-room ESP32 paired windows
(temporal 70/15/15, 70->540 adapter, K=17): pretrained-init 65% PCK@20 vs
scratch 0% (optimization transfer) but frozen-trunk ~0% (no feature
transfer), and NOTHING beats the mean-pose baseline (95.9% PCK@20 —
single subject, near-static normalized coords). Honesty gates held: pred
std 0.0113 (non-constant model) but mean-baseline dominance means no
citable CSI->pose capability from this data. ADR-152 open question 1
answered partially; definitive answer needs multi-subject/position data.

Two new aligner findings: heterogeneous csi_shape with silent zero-padding
(~20%), and extractCsiMatrix's transposed shape label (frame-major data,
[nSc, nFrames] label) — fixes pending.

Co-Authored-By: claude-flow <ruv@ruv.net>

* feat(benchmarks): efficiency sweep MEASURED — half model dominates full reference

Compact WiFlow-STD variants on the same data/split/protocol: half (843,834
params, 0.38x) strictly dominates the 2.23M reference (PCK@20 96.62 vs
96.61, PCK@50 99.47 vs 99.11, MPJPE 0.00898 vs 0.0094) — the published
architecture is over-parameterized for its own benchmark. quarter (338k)
96.05%; tiny (56,290 params, 1/39.5) holds 94.11% — a ~220KB fp32 edge
candidate. In-domain caveats recorded; cross-domain untested.

Co-Authored-By: claude-flow <ruv@ruv.net>

* feat(train): compact WiFlow-STD presets in Rust + tiny edge artifact (ADR-152)

WiFlowStdConfig gains half()/quarter()/tiny() mirroring the overnight sweep
exactly: TcnGroupsMode (Fixed/Gcd/Depthwise), input_pw_groups, derived
stride schedule and decoder-mid (all default to upstream behavior; legacy
serde JSON unaffected). Param formulas pin to trained ground truth first
try: 843,834 / 338,600 / 56,290; default 2,225,042 pin and 1.192e-7 parity
unchanged. 248 tests green.

Tiny edge artifact (tiny_edge_bench.py): ONNX fp32 = 295 KB, 0.66 ms/win
(~1,500/s CPU), 94.11% PCK@20 (matches sweep clean-test exactly; parity
1.49e-7). Static int8 is a bad trade at this scale (-1.43pt, +19% MPJPE,
-16% size, slower) — recorded as negative result. Export note: width-16
breaks AdaptiveAvgPool((15,1)) TorchScript export; replaced by exact
mean+matmul equivalent, proven by parity.

Co-Authored-By: claude-flow <ruv@ruv.net>

* fix: resolve all 10 confirmed code-review findings (7-angle review, 20/20 verified)

wiflow_std: min_feature_width (default 15) replaces the keypoints->stride
coupling — for_keypoints(17) now provably builds the trained [2,2,2,2]
graph and pools 15->17, matching the validated Python protocol (pinned by
tests); param_count() total on invalid configs; random_mask returns Result
and rejects non-finite/out-of-range ratios; trainer checkpoints switched
to safetensors (.pt VarStore roundtrip broken on Windows torch 2.11).

ieee80211bf: SBP proxy now re-triggers instances and relays reports via
Action::RelaySbpReport -> SensingFrame::SbpReport (clients consume via
their existing path); missed_instances reset on success = consecutive
semantics; SessionTable gains a guarded SBP entry point + unknown-id drop
counter; initiator-role sessions reject inbound setup/SBP requests
(RejectedNotSupported) closing the idle hijack; StartSetup/StartSbp
outside Idle return InvalidStateForCommand; SBP validation unified
through evaluate_setup with a 1:1 SetupStatus->SbpStatus mapping.
events.rs split out to honor the 500-line cap.

calibration/cli: enrollment geometry now actually reaches trained banks —
both production call sites attach .with_geometry; --geometry flag on
train-room and POST /enroll/geometry + train-body geometry on
calibrate-serve give production a recording surface; geometry-free banks
log the ADR-152 §2.1.2 note.

benchmarks: corruption masks committed as ground truth (unregenerable
after in-place cleaning; verified bit-identical regeneration from the
pristine copy) + generate_corruption_masks.py producer; _bench_common.py
dedups the 5x-copied shim/evaluate/seed/remap (post-refactor PCK@20
re-verified equal to the last digit); remote scripts get the mmap patch;
tiny_edge --calib validated multiple-of-64; onnx_bench --help no longer
executes (and overwrote) the export — artifact restored byte-exact.

Workspace: 2,963 tests passed, 0 failed; Python proof PASS.

Co-Authored-By: claude-flow <ruv@ruv.net>

* ci: build workspace tests without debuginfo — runner disk exhaustion

The combined 38-crate debug target exceeds the GitHub runner's disk
('final link failed: No space left on device'); the same tree measured
151GB locally with full debuginfo. CARGO_PROFILE_{DEV,TEST}_DEBUG=0
shrinks the target ~5-10x; debuginfo serves no purpose in CI test runs.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-06-11 17:02:23 -04:00

457 lines
18 KiB
Rust
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
//! Full-loop integration test for the ADR-151 calibration pipeline (software half
//! of the §7 validation gap): a clean empty-room **baseline → enroll → extract →
//! train → infer** loop, driven end-to-end through the crates' public API in the
//! exact order the CLI (`calibrate` → `enroll` → `train-room` → `room-watch`)
//! wires the stages.
//!
//! CSI is synthetic but physically plausible:
//! - **empty room**: stable per-subcarrier amplitudes + small complex Gaussian
//! noise (the ADR-135 roundtrip-test fingerprint) — never motion-flagged;
//! - **person present**: a common amplitude offset (extra multipath energy),
//! small body sway, and a constant phase shift. Presence strength is free to
//! exceed z = 2.0 — since the ADR-152 z-band-squeeze fix, anchor motion is
//! measured from frame-to-frame deltas, not from the absolute deviation, so
//! a strongly-reflecting *still* person is no longer misread as "moving";
//! - **breathing**: a few-percent periodic amplitude modulation (0.1250.3 Hz)
//! on a subset of subcarriers — visible in the mean-amplitude scalar the CLI
//! uses, invisible to the per-frame *median* z (so still anchors stay still);
//! - **small movement**: per-frame amplitude jitter + a phase wobble that swings
//! past the π/6 drift threshold.
//!
//! Deterministic (xorshift32, fixed seeds), no I/O, no hardware. What remains
//! hardware-only is the on-target run with real ESP32 CSI and a live operator.
use std::f32::consts::PI;
use ndarray::Array2;
use num_complex::Complex64;
use wifi_densepose_calibration::extract::Features;
use wifi_densepose_calibration::{
AnchorFeature, AnchorLabel, AnchorQualityGate, AnchorRecorder, EnrollmentEvent,
EnrollmentSession, MixtureOfSpecialists, NodeGeometry, SpecialistBank, SpecialistKind,
};
use wifi_densepose_core::types::{AntennaConfig, CsiFrame, CsiMetadata, DeviceId, FrequencyBand};
use wifi_densepose_signal::{BaselineCalibration, CalibrationConfig, CalibrationRecorder};
// ---------------------------------------------------------------------------
// Deterministic PRNG (xorshift32 + Box-Muller) — same pattern as
// wifi-densepose-signal/tests/calibration_roundtrip.rs.
// ---------------------------------------------------------------------------
struct Rng(u32);
impl Rng {
fn new(seed: u32) -> Self {
assert_ne!(seed, 0, "xorshift seed must be non-zero");
Self(seed)
}
fn next_u32(&mut self) -> u32 {
let mut x = self.0;
x ^= x << 13;
x ^= x >> 17;
x ^= x << 5;
self.0 = x;
x
}
fn next_normal(&mut self) -> f32 {
let u1 = (self.next_u32() as f32 + 1.0) / (u32::MAX as f32 + 2.0);
let u2 = (self.next_u32() as f32 + 1.0) / (u32::MAX as f32 + 2.0);
(-2.0 * u1.ln()).sqrt() * (2.0 * PI * u2).cos()
}
}
// ---------------------------------------------------------------------------
// Synthetic room (HT20: 52 active subcarriers @ 20 Hz)
// ---------------------------------------------------------------------------
const N_SC: usize = 52;
const FS_HZ: f32 = 20.0;
/// Complex-noise std per quadrature ⇒ amplitude noise std ≈ NOISE_STD.
const NOISE_STD: f32 = 0.01;
/// Capture length per enrollment anchor (20 s @ 20 Hz; gate needs ≥ 60).
const ANCHOR_FRAMES: usize = 400;
/// Baseline / runtime window length (30 s @ 20 Hz; recorder needs ≥ 600).
const WINDOW_FRAMES: usize = 600;
/// What the person in the room is doing (None ⇒ empty room).
#[derive(Clone, Copy, Default)]
struct Person {
/// Common amplitude offset in units of NOISE_STD (presence strength).
/// Anything ≥ 1.5 reads as present; values above 2.0 are explicitly
/// exercised to guard the ADR-152 z-band-squeeze fix (presence strength
/// must not read as motion).
presence_z: f32,
/// Per-frame common amplitude jitter (body sway / fidgeting), in NOISE_STD.
sway_z: f32,
/// Respiration rate (Hz); 0 = no modulation.
breathing_hz: f32,
/// Relative amplitude-modulation depth on every 4th subcarrier.
breathing_depth: f32,
/// Constant phase shift from the body's multipath (radians).
phase_shift: f32,
/// Phase-wobble amplitude (radians) at 1.5 Hz — drives the motion flag.
phase_wobble: f32,
}
/// Deterministic CSI source for one room. Time advances one frame per call.
struct RoomSim {
rng: Rng,
/// Static per-subcarrier amplitude fingerprint.
amp: Vec<f32>,
/// Static per-subcarrier phase fingerprint.
phase: Vec<f32>,
/// Frame counter (continuous room clock).
t: u64,
}
impl RoomSim {
fn new(seed: u32) -> Self {
// Same HT20 fingerprint as the ADR-135 roundtrip test.
let amp = (0..N_SC)
.map(|k| 0.3 + 0.7 * (k as f32 * PI / N_SC as f32).sin().abs())
.collect();
let phase = (0..N_SC)
.map(|k| (k as f32 * 0.1).rem_euclid(2.0 * PI) - PI)
.collect();
Self { rng: Rng::new(seed), amp, phase, t: 0 }
}
/// Generate the next CSI frame for the given occupancy.
fn frame(&mut self, person: Option<&Person>) -> CsiFrame {
let secs = self.t as f32 / FS_HZ;
let (offset, wobble) = match person {
Some(p) => {
let sway = p.sway_z * NOISE_STD * self.rng.next_normal();
(
p.presence_z * NOISE_STD + sway,
p.phase_shift + p.phase_wobble * (2.0 * PI * 1.5 * secs).sin(),
)
}
None => (0.0, 0.0),
};
let mut data = Array2::<Complex64>::zeros((1, N_SC));
for k in 0..N_SC {
let mut a = self.amp[k] + offset;
if let Some(p) = person {
if p.breathing_hz > 0.0 && k % 4 == 0 {
a *= 1.0 + p.breathing_depth * (2.0 * PI * p.breathing_hz * secs).sin();
}
}
let th = self.phase[k] + wobble;
let re = a * th.cos() + NOISE_STD * self.rng.next_normal();
let im = a * th.sin() + NOISE_STD * self.rng.next_normal();
data[(0, k)] = Complex64::new(re as f64, im as f64);
}
let mut meta =
CsiMetadata::new(DeviceId::new("full-loop-test"), FrequencyBand::Band2_4GHz, 6);
meta.bandwidth_mhz = 20;
meta.antenna_config = AntennaConfig::new(1, 1);
self.t += 1;
CsiFrame::new(meta, data)
}
}
/// Per-frame scalar — mean amplitude across subcarriers/streams, the same
/// carrier the CLI's `frame_scalar` feeds into `Features::from_series`.
fn frame_scalar(frame: &CsiFrame) -> f32 {
frame.mean_amplitude() as f32
}
/// Synthetic occupancy for each guided anchor in the canonical sequence.
fn anchor_person(label: AnchorLabel) -> Option<Person> {
let p = match label {
AnchorLabel::Empty => return None,
// Strong reflector at z = 3.0 — every frame exceeds the baseline's
// absolute motion threshold (z > 2.0). Pre-ADR-152 this anchor was
// unenrollable ("too much motion"); the delta-based gate must accept it.
AnchorLabel::StandStill => Person {
presence_z: 3.0, sway_z: 0.25, phase_shift: 0.10, ..Default::default()
},
AnchorLabel::Sit => Person {
presence_z: 1.65, sway_z: 0.25, phase_shift: 0.08, ..Default::default()
},
AnchorLabel::LieDown => Person {
presence_z: 1.6, sway_z: 0.25, phase_shift: 0.06, ..Default::default()
},
AnchorLabel::BreatheSlow => Person {
presence_z: 1.7, sway_z: 0.2, breathing_hz: 0.125, breathing_depth: 0.03,
phase_shift: 0.08, ..Default::default()
},
AnchorLabel::BreatheNormal => Person {
presence_z: 1.7, sway_z: 0.2, breathing_hz: 0.25, breathing_depth: 0.03,
phase_shift: 0.08, ..Default::default()
},
AnchorLabel::SmallMove => Person {
presence_z: 1.7, sway_z: 1.0, phase_shift: 0.10, phase_wobble: 1.0,
..Default::default()
},
AnchorLabel::SleepPosture => Person {
presence_z: 1.6, sway_z: 0.2, breathing_hz: 0.2, breathing_depth: 0.03,
phase_shift: 0.06, ..Default::default()
},
};
Some(p)
}
/// Capture one anchor exactly as the CLI's `enroll` does: per-frame deviation
/// into the `AnchorRecorder`, scalar series for feature extraction, then the
/// quality-gate verdict.
fn capture_anchor(
sim: &mut RoomSim,
baseline: &BaselineCalibration,
gate: &AnchorQualityGate,
label: AnchorLabel,
room_id: &str,
at_unix_s: i64,
) -> (Option<AnchorFeature>, wifi_densepose_calibration::Anchor, Option<String>) {
let person = anchor_person(label);
let mut recorder = AnchorRecorder::new(label);
let mut series = Vec::with_capacity(ANCHOR_FRAMES);
for _ in 0..ANCHOR_FRAMES {
let frame = sim.frame(person.as_ref());
recorder.record_frame(baseline, &frame);
series.push(frame_scalar(&frame));
}
let (anchor, reason) = recorder.finalize(gate, at_unix_s);
let feature = anchor
.quality
.accepted
.then(|| AnchorFeature::from_series(room_id, label, &series, FS_HZ));
(feature, anchor, reason)
}
/// Generate a live feature window (Stage-5 runtime input).
fn live_window(sim: &mut RoomSim, person: Option<&Person>) -> Features {
let series: Vec<f32> = (0..WINDOW_FRAMES)
.map(|_| frame_scalar(&sim.frame(person)))
.collect();
Features::from_series(&series, FS_HZ)
}
// ---------------------------------------------------------------------------
// The full loop
// ---------------------------------------------------------------------------
#[test]
fn full_loop_baseline_enroll_extract_train_infer() {
let room_id = "living-room";
let mut sim = RoomSim::new(42);
// -- Stage 1: clean empty-room baseline capture (ADR-135) ----------------
let mut recorder = CalibrationRecorder::new(CalibrationConfig::ht20());
let mut flagged_after_warmup = 0u32;
for i in 0..WINDOW_FRAMES {
let frame = sim.frame(None);
let score = recorder.record(&frame).expect("baseline record");
// Welford stats need a short warmup before the partial z is meaningful.
if i >= 100 && score.motion_flagged {
flagged_after_warmup += 1;
}
}
assert_eq!(recorder.frames_recorded(), WINDOW_FRAMES as u32);
assert_eq!(
flagged_after_warmup, 0,
"a static empty room must never be motion-flagged after warmup"
);
let baseline = recorder.finalize().expect("baseline finalize");
assert_eq!(baseline.subcarriers.len(), N_SC);
let baseline_id = baseline.calibration_uuid().to_string();
// A fresh empty frame deviates negligibly from its own baseline.
let check = baseline.deviation(&sim.frame(None)).expect("deviation");
assert!(!check.motion_flagged, "empty frame flagged: {check:?}");
assert!(
check.amplitude_z_median < 1.0,
"empty frame z {} should be < 1.0",
check.amplitude_z_median
);
// -- Stage 2: guided-anchor enrollment with the quality gate -------------
let gate = AnchorQualityGate::default();
let mut session = EnrollmentSession::new(room_id, &baseline_id, 1_700_000_000);
// Transceiver geometry recorded at session start (ADR-152 §2.1.1): a
// two-node layout, one tape-measured, one unknown — all fields optional.
let geometry = vec![
NodeGeometry::new(1, "tape-measure")
.with_position(0.0, 0.0, 1.2)
.with_orientation(0.0, 0.0)
.with_distance(2, 3.5),
NodeGeometry::unknown(2),
];
session.record_geometry(geometry.clone(), 1_700_000_000);
assert_eq!(session.geometry(), Some(geometry.as_slice()));
let mut features: Vec<AnchorFeature> = Vec::new();
for (i, label) in AnchorLabel::SEQUENCE.into_iter().enumerate() {
let at = 1_700_000_000 + (i as i64 + 1) * 30;
let (feat, anchor, reason) =
capture_anchor(&mut sim, &baseline, &gate, label, room_id, at);
assert!(
anchor.quality.accepted,
"anchor {} rejected: {} (presence_z={:.2} motion={:.0}% frames={})",
label.as_str(),
reason.unwrap_or_default(),
anchor.quality.presence_z,
anchor.quality.motion_rate * 100.0,
anchor.quality.frames,
);
match label {
AnchorLabel::Empty => assert!(
anchor.quality.presence_z < 1.0,
"empty room must read empty, got z {}",
anchor.quality.presence_z
),
AnchorLabel::SmallMove => assert!(
anchor.quality.motion_rate >= 0.3,
"small-move motion {} too low",
anchor.quality.motion_rate
),
_ => assert!(
anchor.quality.presence_z >= 1.5,
"{} presence_z {} below gate",
label.as_str(),
anchor.quality.presence_z
),
}
features.push(feat.expect("accepted anchor yields a feature"));
session.apply(EnrollmentEvent::AnchorAccepted { anchor });
}
assert!(session.is_complete(), "missing anchors: {:?}", session.missing());
assert_eq!(session.progress(), (8, 8));
session.apply(EnrollmentEvent::Completed { at: 1_700_000_300 });
// -- Stage 3: feature extraction sanity ----------------------------------
assert_eq!(features.len(), 8);
let by_label = |l: AnchorLabel| {
features
.iter()
.find(|f| f.label == l)
.unwrap_or_else(|| panic!("no feature for {}", l.as_str()))
};
let breathe = by_label(AnchorLabel::BreatheNormal);
assert!(
(breathe.features.breathing_hz - 0.25).abs() < 0.04,
"normal breathing extracted at {} Hz, injected 0.25 Hz",
breathe.features.breathing_hz
);
assert!(
breathe.features.breathing_score > 0.25,
"breathing score {} too weak",
breathe.features.breathing_score
);
let slow = by_label(AnchorLabel::BreatheSlow);
assert!(
(slow.features.breathing_hz - 0.125).abs() < 0.04,
"slow breathing extracted at {} Hz, injected 0.125 Hz",
slow.features.breathing_hz
);
let empty = by_label(AnchorLabel::Empty);
assert!(
empty.features.variance < breathe.features.variance,
"empty variance {} should be below occupied {}",
empty.features.variance,
breathe.features.variance
);
// -- Stage 4: train the specialist bank + JSON persistence round-trip ----
// The bank snapshots the geometry the enrollment recorded (ADR-152 §2.1.1).
let bank = SpecialistBank::train(room_id, &baseline_id, &features, 1_700_000_400)
.expect("bank training")
.with_geometry(session.geometry().map(<[_]>::to_vec).unwrap_or_default());
assert_eq!(bank.room_id, room_id);
assert_eq!(bank.anchor_count, 8);
let kinds = bank.trained_kinds();
for kind in [
SpecialistKind::Presence,
SpecialistKind::Posture,
SpecialistKind::Breathing,
SpecialistKind::Heartbeat,
SpecialistKind::Restlessness,
SpecialistKind::Anomaly,
] {
assert!(kinds.contains(&kind), "bank missing {kind:?} (got {kinds:?})");
}
// Persist and reload (JSON today) — the runtime below uses the *reloaded*
// bank, so the round-trip is proven inside the loop, not as a side check.
let json = bank.to_json().expect("bank to_json");
let reloaded = SpecialistBank::from_json(&json).expect("bank from_json");
assert_eq!(reloaded.room_id, bank.room_id);
assert_eq!(reloaded.baseline_id, bank.baseline_id);
assert_eq!(reloaded.anchor_count, bank.anchor_count);
assert_eq!(
reloaded.presence.as_ref().map(|p| p.threshold),
bank.presence.as_ref().map(|p| p.threshold),
"presence threshold must survive persistence"
);
assert_eq!(
reloaded.geometry, geometry,
"the enrollment geometry snapshot must survive bank persistence"
);
// -- Stage 5: runtime inference through the mixture ----------------------
let mix = MixtureOfSpecialists::new(reloaded);
// Positive case: a person breathing at a KNOWN 0.30 Hz (18 BPM) — a rate
// never used during enrollment.
let occupied = Person {
presence_z: 1.7,
sway_z: 0.25,
breathing_hz: 0.30,
breathing_depth: 0.04,
phase_shift: 0.08,
..Default::default()
};
let f = live_window(&mut sim, Some(&occupied));
let state = mix.infer(&f, &baseline_id);
assert!(!state.stale, "bank trained against this baseline must be fresh");
assert!(!state.vetoed, "plausible occupied window must not be vetoed");
let presence = state.presence.expect("presence specialist trained");
assert_eq!(presence.value, 1.0, "person in the room must be detected");
let breathing = state.breathing.expect("breathing must be reported when present");
assert!(
(breathing.value - 18.0).abs() <= 2.0,
"breathing {} BPM, injected 18 BPM",
breathing.value
);
assert!(state.restlessness.is_some(), "restlessness specialist trained");
// Motionless-person case (ADR-152 "variance-only presence" regression):
// a strong reflector standing perfectly still — variance stays at the
// empty-room level, only the scalar MEAN shifts. The mean channel of the
// presence specialist must still detect them.
let motionless = Person {
presence_z: 3.0,
sway_z: 0.05,
phase_shift: 0.10,
..Default::default()
};
let f_still = live_window(&mut sim, Some(&motionless));
let state = mix.infer(&f_still, &baseline_id);
let presence = state.presence.expect("presence specialist trained");
assert_eq!(
presence.value, 1.0,
"motionless person must be detected via the mean-shift channel \
(variance {:.2e} vs empty-level)",
f_still.variance
);
// Negative case: a fresh empty-room window must NOT report presence,
// breathing, heartbeat, or posture.
let f_empty = live_window(&mut sim, None);
let state = mix.infer(&f_empty, &baseline_id);
let presence = state.presence.expect("presence specialist trained");
assert_eq!(presence.value, 0.0, "empty room must read absent");
assert!(state.breathing.is_none(), "no breathing in an empty room");
assert!(state.heartbeat.is_none(), "no heartbeat in an empty room");
assert!(state.posture.is_none(), "no posture in an empty room");
// Honest degradation: a drifted baseline flags the bank STALE.
let state = mix.infer(&f, "some-other-baseline");
assert!(state.stale, "baseline drift must mark readings STALE");
}