Spencer Morris スペンサー・モリス← the streetLet's talk
SM28

SPECTRE

CASE STUDY · 2 MIN
Execution / case study

A passive RF sensing network from edge to cloud

Edge receivers classify what they hear with a CNN that beats its baseline on receivers it never trained on. A signed pipeline carries every detection to the cloud. A 384-receiver simulation shows that GPS-disciplined clocks bring position fixes to within 6 m.

ExecutionEdge-to-cloud RF pipeline, lab and simulationNov 2026Architecture, DSP, classifier, localization study, console
CNN closed-set accuracy on held-out receivers, vs 0.934 for a gradient-boosted baseline
median TDOA error with 12 GPS-disciplined sensors vs NTP clocks
virtual receivers in the local fleet
parameter CNN, shipped as a signed model only when it beats the baseline
1 SPECTRE

Measured against what the hardware can do

Edge agents run a DSP chain on RTL-SDR-style I/Q (adaptive noise floor, detection, features, an open-set 'unknown' class) and upload Ed25519-signed batches through a gateway with replay windows, quarantine and backpressure, into SQS, a batched idempotent worker and Postgres. A 51,802-parameter CNN trained on impaired simulated I/Q scored 0.976 accuracy and 0.973 macro-F1 on receivers it never saw in training, and only ships as a signed ONNX model if it beats the baseline's macro-F1 by 0.02.

ADR-0005 makes the live system report localization as unavailable: NTP gives millisecond timing and TDOA needs about 10 ns. The localization engine (RSSI, TDOA, angle of arrival, joint solves, a Kalman tracker) runs only on simulated measurements. A 120-trial Monte Carlo per sensor count puts numbers on the gap: with 12 GPS-disciplined sensors TDOA lands within 6 m median, with NTP clocks it misses by 6.2 km. Results are labelled simulated throughout, because synthetic data overstates accuracy.

⏱ Timetable

Timeline

  1. Backend, simulator, edge agent, gateway, worker, query API, compose stack
  2. Signed CNN classifier and localization engine
  3. Air picture: terrain, probability fields, TDOA hyperboloids
PythonPyTorchONNXPostgresSQS (ElasticMQ)S3 (SeaweedFS)Docker ComposeReact
Sources and method

Written from the project's code, documentation and commit history, reviewed in November 2026. Business figures come from me. Clips and screenshots are labelled with where they came from; none show customer data.

  • spectre/README.md
  • spectre/docs/ROADMAP.md
  • spectre/docs/ml.md
  • spectre/docs/localization.md
  • spectre/docs/adr/0005-honest-localization.md
  • spectre/models/gate.json