[Research] Ghost Murmur on RuView — open-source spec for an honest multi-modal heartbeat mesh

[ruvnet]

TL;DR

In early April 2026 the CIA reportedly used a Lockheed Skunk Works system called "Ghost Murmur" — NV-diamond quantum magnetometry + AI — to help locate downed F-15E pilot "Dude 44 Bravo" in southern Iran. President Trump publicly suggested detection ranges as long as 40 miles. Physicists across Scientific American, Military.com, and others pushed back: the heart's magnetic field falls off as ~1/r³ and a multi-mile single-heartbeat detection in an uncontrolled outdoor environment is not consistent with publicly documented physics.

This issue tracks a research spec that does two things:

1. Reality-checks the news against peer-reviewed physics.
2. Maps a sober version onto RuView — we already ship ~80% of the building blocks for an honestly-scoped multi-modal heartbeat mesh.

Spec doc: docs/research/quantum-sensing/16-ghost-murmur-ruview-spec.md (583 lines, branch feat/nvsim-pipeline-simulator)
Public gist: https://gist.github.com/ruvnet/e44d0c3f0ad10d9c4933a196a16d405c

---

Why this matters

RuView is open-source civilian sensing — occupancy monitoring, vital signs, search-and-rescue, mass-casualty triage. The press around Ghost Murmur asks the same physical question we have been answering for two years: can a network of cheap sensors find and track a live human at meaningful range? The answer in the open literature is: yes, at room-and-building scale; no, at multi-mile scale. This issue collects the evidence and points at the existing crates that already do the room-scale version honestly.

---

What was claimed (per the public press)

Claim	Source
NV defects in synthetic diamond as the sensor	All outlets
AI strips environmental noise to isolate cardiac signal	All outlets
Operates at room temperature in smaller packages than SQUIDs	Military.com
Detection range "tens of miles"	Trump remarks, WION, Open The Magazine
Developed by Lockheed Martin Skunk Works	All outlets
First operational use in Apr 2026 Iran rescue	Newsweek, Yahoo

---

What the physics actually allows

Heart's magnetocardiogram (MCG) signal vs. distance, dipole-approximation falloff:

Distance	Peak QRS field
10 cm	50 pT
1 m	50 fT
10 m	50 aT
1 km	5 × 10⁻²³ T
40 mi (65 km)	~10⁻²⁸ T

For reference: Earth's field is ~50 µT (a billion × the chest signal at 10 cm); the best published NV ensemble laboratory record is 0.9 pT/√Hz [Wolf 2015]; a shielded SQUID reaches ~1 fT/√Hz. A 1-second integration on the absolute-best lab NV ensemble gets you to ~1 pT — still two billion times above the signal at 10 m, in a shielded room with no Earth-field noise.

Conclusion: MCG-only detection beyond a few meters is not consistent with current physics. If the Ghost Murmur story is even partially real, the most defensible interpretation is close-approach gradiometric MCG to confirm a heat-signature is alive at meters from a low-hovering platform — not multi-mile detection. Most of the actual rescue almost certainly came from a survival beacon (PRC-112/CSEL on 121.5/243/406 MHz + UHF SATCOM) plus IR thermal + mmWave from low-flying ISR — none of which need a quantum sensor.

---

RuView mapping: three-tier mesh (already shipping)

                      ┌──────────────────────────┐
                      │   Tier 3 — NV-diamond    │  Range: 0.1–2 m (today, lab)
                      │     magnetometer ring    │  Status: nvsim simulator only
                      └──────────┬───────────────┘
                      ┌──────────┴───────────────┐
                      │   Tier 2 — 60 GHz FMCW   │  Range: 1–10 m HR/BR
                      │     mmWave radar mesh    │  Status: shipping (ADR-021)
                      └──────────┬───────────────┘
                      ┌──────────┴───────────────┐
                      │  Tier 1 — WiFi CSI mesh  │  Range: 10–30 m through-wall
                      │   (presence, breathing,  │  Status: shipping (ADR-014, ADR-029)
                      │    pose, intention)      │  Hardware: $9 ESP32-S3 8MB
                      └──────────┬───────────────┘
                                 ▼
                  ┌────────────────────────────────┐
                  │  RuvSense multistatic fusion   │
                  │   + cross-viewpoint attention  │
                  │   + AETHER re-ID embeddings    │
                  │   + Cramer-Rao gating          │
                  └────────────────────────────────┘

Press claim	RuView equivalent	Crate / ADR	Honest range
NV-diamond magnetometry	Deterministic NV pipeline simulator	nvsim / ADR-089	Simulator (no hw yet)
AI noise stripping	RuvSense multistatic fusion + AETHER	signal/ruvsense/, ADR-029, ADR-024	Mature
Heartbeat at distance	60 GHz FMCW HR/BR + WiFi CSI breathing	wifi-densepose-vitals (ADR-021), signal	1–5 m HR; 10–30 m presence
Long-range localisation	Multistatic time-of-flight + CRLB	ruvector/viewpoint/geometry.rs	Limited by node spacing

---

Concrete build (today, $165)

Tier	Component	Qty	Per-unit
1	ESP32-S3 8 MB DevKit	3	$9
1	PoE injector + cat6	3	$6
2	ESP32-C6 + Seeed MR60BHA2	1	$15
3	(NV node — nvsim only)	0	—
Edge	Pi 5 (8 GB) or mini-PC	1	$80
Net	unmanaged GbE switch	1	$25
Total			$165

Honest performance you can expect on this build:

Metric	Expected (3 CSI + 1 mmW + nvsim row)
Person detection (LOS)	95% TPR / 5% FPR @ 0–15 m
Person detection (1 wall)	85% TPR / 8% FPR @ 0–10 m
HR accuracy (LOS, 0–3 m)	±2 bpm
BR accuracy (any mode)	±1 br/min
Pose keypoint error (LOS)	~10 cm @ 0–5 m
Latency (sensor → fused)	80–150 ms

This is not 40 miles. It is a small house. That is the entire point.

Full step-by-step build (firmware flash, mesh.toml, web UI launch) in §10 of the spec.

---

Privacy, ethics, legal

A heartbeat-detecting mesh is dual-use. RuView's project line:

1. Civilian, opt-in deployments only — SAR, elder-care, occupancy, ICU vitals. Not surveillance.
2. No directional pursuit — no beam-steering or remote person-of-interest tracking primitives.
3. Data minimisation — fused output (presence, HR, BR, pose, p_alive) only; raw streams discarded at the edge.
4. PII gates (ADR-040) block identifying biometric streams from leaving the local mesh without consent.
5. Adversarial-signal detection (ruvsense::adversarial) protects against malicious mesh nodes injecting false detections.
6. No export-controlled hardware — RuView targets <$50 COTS; ITAR/EAR sub-THz coherent radars and shielded NV ensembles are explicitly out of scope.

The Ghost Murmur press story exists in a different ethical universe — covert military intel ops with no consent, no notice, no opt-out. RuView is not that. This spec is the open-source version: same physics, opposite governance.

---

Comparison table

Dimension	RuView heartbeat mesh	Press-claimed Ghost Murmur
Range	0.5–30 m	tens of miles
Modalities	WiFi CSI + 60 GHz radar + NV simulator	NV-diamond only (per press)
Cost per node	$9–15	unstated
Through-wall	yes (CSI)	unstated
Vital signs	HR + BR + pose	claimed: HR
Open source	yes (Apache-2.0 / MIT)	classified
Independent verification	yes (SHA-256 witnesses, ADR-028)	no
Plausible per published physics	yes	not at the claimed ranges
Ethics governance	civilian opt-in	covert military
Build today on $200	yes	no

---

Open research questions (PRs welcome)

1. Sub-$1k NV-ensemble board (rumored at QDM Tech, NVision, Adamas — nothing shipping)
2. FCC-compliant 76–81 GHz cardiac stand-off radar
3. Distributed coherent processing across CSI nodes (true multistatic phase-coherent SAR — needs sub-ns clock sync)
4. RaBitQ binary-sketch novelty gate on ESP32 firmware (ADR-086)
5. Differential-privacy fusion for the heartbeat-mesh output stream
6. nvsim validated against a published cardiac-magnetic-field reference (currently internally consistent only)

---

Cross-references

• ADR-014 — SOTA signal processing
• ADR-021 — ESP32 CSI-grade vital-sign extraction
• ADR-024 — AETHER contrastive embedding
• ADR-027 — MERIDIAN cross-environment domain generalisation
• ADR-028 — ESP32 capability audit + witness verification
• ADR-029 — RuvSense multistatic sensing mode
• ADR-040 — PII detection gates
• ADR-086 — ESP32-side novelty gate (RaBitQ)
• ADR-089 — nvsim NV-diamond pipeline simulator (Accepted)
• ADR-090 — nvsim Lindblad/Hamiltonian extension (Proposed, conditional)

---

References

Primary physics

• Cohen 1970, Appl. Phys. Lett. 16, 278 — first MCG measurement
• Bison 2009, Appl. Phys. Lett. 95, 173701 — room-temperature 19-channel MCG mapper
• Wolf 2015, Phys. Rev. X 5, 041001 — subpicotesla diamond magnetometry
• Barry 2020, Rev. Mod. Phys. 92, 015004 — NV magnetometry sensitivity (the proxy validity reference for nvsim)
• Doherty 2013, Phys. Rep. 528, 1–45 — NV centre review
• Jackson 3e §5.6, §5.8 — dipole + Biot-Savart

News (April 2026)

• Newsweek — What Is Ghost Murmur?
• Scientific American — Scientists question CIA's claim
• Military.com — Heartbeat-Tracking Tech Has Experts Questioning Laws of Physics
• Open The Magazine — Inside CIA's Chilling New Tech
• Calcalist — Spy tech or science fiction?
• HN thread #47679241

---

Status / acceptance

• Spec doc committed to feat/nvsim-pipeline-simulator
• Published as public gist
• Cross-referenced from existing ADRs (089, 090)
• Reviewed by maintainer
• Merged to main alongside PR feat(nvsim): full simulator stack — Rust crate, dashboard, server, App Store, Ghost Murmur [ADR-089/090/091/092/093] #436

RuView is an open-source civilian sensing platform. It is not affiliated with the United States government, the CIA, Lockheed Martin, or any classified program. References to "Ghost Murmur" in this document refer exclusively to the publicly-reported program of that name as covered in the open press in April 2026.

[ruvnet]

Stand-off radar tier (77 GHz high-power + 100-200 GHz coherent sub-THz) decision now documented as ADR-091 — Proposed, research only, no production hardware. Decision matrix: skip permanently for hardware in 77 GHz beyond §95.M / 220 GHz / 380+ GHz; research-only simulator-class artifact for 77 GHz at §95.M, 100 GHz, 140 GHz; build now: nothing. Mirrors the nvsim conditional-build pattern (ADR-089/090). Commit 508e2c6 on feat/nvsim-pipeline-simulator.

[ruvnet]

Status update — full implementation merged

PR #436 merged at squash 7f5a69263 on main (2026-04-27).

The Ghost Murmur research is now backed by a working, deployed, open-source implementation:

Live

• Dashboard: https://ruvnet.github.io/RuView/nvsim/
• Ghost Murmur view in dashboard: navigate to the ghost-icon in the rail
• Live demo: distance/moment sliders drive the real Rust nvsim pipeline via WASM. At 10 cm with cardiac-strength dipole the recovered |B| floors at the ADC quantization residual (~2 nT) — vindicating the spec's central physics teaching.

Code

• nvsim crate (ADR-089 Accepted) — v2/crates/nvsim. Deterministic NV-diamond pipeline, 50/50 native tests, pinned reference witness cc8de9b01b0ff5bd97a6c17848a3f156c174ea7589d0888164a441584ec593b4.
• nvsim-server crate (ADR-092 §6.2) — v2/crates/nvsim-server. Axum REST + WS host. Dockerfile + ghcr.io publish workflow.
• Dashboard (ADR-092 Implemented) — dashboard/. Vite + Lit + WASM transport + WS transport.
• Ghost Murmur view — dashboard/src/components/nv-ghost-murmur.ts. Full research spec rendered + interactive demo.
• Spec doc — docs/research/quantum-sensing/16-ghost-murmur-ruview-spec.md (583 lines).

Three-tier mesh — all three transports

• Tier 1 — WiFi CSI (10–30 m through-wall): wifi-densepose-signal + wifi-densepose-wasm-edge (75/75 tests, ESP32 deployment ready)
• Tier 2 — 60 GHz mmWave (1–10 m HR/BR): wifi-densepose-vitals, ADR-021
• Tier 3 — NV-diamond (0.1–2 m sim, hardware deferred): nvsim, ADR-089

CI / infra

• .github/workflows/dashboard-pages.yml — auto-deploys to gh-pages/nvsim/
• .github/workflows/dashboard-a11y.yml — axe-core × {Chromium, Firefox, WebKit}
• .github/workflows/nvsim-server-docker.yml — publishes ghcr.io/ruvnet/nvsim-server

Honest gaps remaining

• §11.5 axe-core formal scan — workflow added, first run pending
• §11.8 Cross-browser — workflow added, first run pending
• §11.12 WS transport byte-equivalence — nvsim-server Docker image built, public deploy pending
• 59 of 65 wasm-edge apps still need ESP32 mesh + WS transport for live runtime; 6 ship a simulated runtime in-browser

The Ghost Murmur reality-check thesis stands: the press claim of 40-mile cardiac MCG sits 12+ orders of magnitude below any published instrument's noise floor. Open the live demo, slide the distance from 10 cm out toward 1 km, and watch the per-tier detectability bars flip to "× too weak" exactly as Wikswo / Orzel / Roth predicted in the Scientific American skeptic's piece.