Skip to content
Vasilis edited this page Nov 6, 2026 · 3 revisions

SpectrumAlert Wiki

RF spectrum monitoring, anomaly detection, transmitter geolocation and device identification.

SpectrumAlert dashboard

SpectrumAlert is an operator console for RTL-SDR receivers. It provides a web dashboard, a terminal interface, headless services and a Docker deployment. All interfaces share one detector, one storage format and one data directory.

Start here

Page Purpose
Installation Local install, SDR permissions, Docker
Quick Start First dashboard session and first scan
Web Dashboard Modules, workflow and screenshots
Triangulation Multi-receiver locating, fit quality and residuals
CLI Reference Terminal commands and options
Docker Compose services, USB access, ports
Data and Training Capture storage, IQ fixtures, model training
Troubleshooting Receiver, port and USB problems
FAQ GUI vs CLI, dBFS, ports, common questions

At a glance

  • Live spectrum, waterfall and detection timeline
  • SNR and prominence based anomaly detection
  • Multi-receiver triangulation with confidence grading
  • Per-receiver calibration and path-loss fitting
  • RF fingerprinting with a reference library
  • Remote rtl_tcp receivers
  • PMR446 channel scanner
  • Model training, testing and deployment
  • Dark and light themes, responsive layout
  • Offline-capable UI assets
  • Deterministic synthetic IQ fixtures for development

GUI and CLI are both supported

The web dashboard is not a replacement for the terminal. It is another front end to the same backend:

# Terminal
spectrum-alert monitor status
spectrum-alert monitor start --freq-start 144 --freq-end 148

# Web dashboard
spectrum-alert web --port 8000

# Docker dashboard
docker compose up -d

# Docker CLI command
docker compose run --rm spectrumalert monitor status

Only one process can hold a physical RTL-SDR at a time. Stop dashboard monitoring before running a CLI scan against the same dongle.

Signal levels

Levels are dBFS, not dBm. The RTL-SDR has no absolute power reference until a per-receiver calibration offset is applied. See Data and Training for the storage model.

Clone this wiki locally