Back to RTG Spectrum

RF Sensing

RF sensing and signal classification software development.

RTG Spectrum develops RF signal classification software for passive detection, SDR capture, spectrum monitoring, RFML experimentation, edge inference, and operator-facing RF analysis. The work spans signal feature extraction, known-vs-unknown triage, confidence scoring, replay testing, event normalization, Pattern-of-Life analytics, and multi-sensor correlation across RF, WiFi, BLE, cellular, and GPS data.

RF Sensing Engineering Geospatial RF Analytics

Service Focus

RF signal detection, classification, and analysis from collection to operator workflow.

RTG Spectrum provides RF sensing platform engineering for collection paths, signal-processing pipelines, event normalization, ML-assisted classification, dashboard workflows, exports, and technical evidence for partner evaluation. Trace Analyzer shows how BLE, GPS, WiFi, Cellular Survey, SDR-Shark, RF Sentinel, and PASSIVE-SHIELD passive DJI/OcuSync evidence can share one reviewable timeline.

Common Needs

  • Passive RF detection and signal classification software
  • RF feature extraction, modulation or signal-family classification, confidence scoring, and threshold tuning
  • Operator UI, event streams, exports, and reporting
  • Integration with edge compute, SDRs, and sensor systems
  • Cellular survey, RF, WiFi, BLE, GPS, and mission cues in one operator timeline

Signal Classification

RF signal classification software for SDR and edge sensing workflows.

RF classification workflows can combine signal detection, feature extraction, deterministic signal-family rules, and ML-assisted inference to help operators distinguish and prioritize wireless activity. RTG Spectrum develops these workflows around SDR and edge-compute platforms, including collection, preprocessing, classification, confidence scoring, event storage, operator visualization, and replay fixtures for repeatable evaluation.

Classification Workflow

  • SDR signal analysis and spectrum monitoring inputs
  • Signal feature extraction, confidence scoring, and RF analytics
  • Known-signal matching, unknown-signal triage, and source-held-out evaluation planning
  • Automatic modulation or signal-family classification support where evidence allows
  • Source-aware events that can feed Trace Analyzer and Pattern-of-Life review

Evaluation Discipline

Classification claims need replay data, confidence, and held-out tests.

RTG Spectrum treats classification as an evidence workflow, not just a model output. A useful classifier needs labeled or explainable inputs, repeatable replay, confidence and threshold behavior, known-vs-unknown handling, source-held-out evaluation where possible, and operator-visible context so a reviewer can understand why the software raised a cue.

Evidence Outputs

  • Replayable IQ, feature, or event fixtures for regression testing
  • Classifier confidence, threshold, and false-positive review notes
  • Known-family and unknown-family handling boundaries
  • Operator screenshots, event exports, and Trace Analyzer context

Demonstrated and experimental paths

Clear maturity boundaries make RF classification software more credible.

Demonstrated Public Artifacts

SDR-Shark shows replay-backed spectrum and waterfall workflows. RF Sentinel shows normalized Bluetooth Classic, BLE, Zigbee, and WiFi detections, entity tracking, topology, Pattern-of-Life, and shared event outputs.

View RF Sentinel

Mission-Oriented Evidence

PASSIVE-SHIELD shows passive DJI/OcuSync-family prototype evidence, acoustic corroboration, deterministic swarm replay, GeoFusion residual analysis, and bounded hypothesis outputs.

Read PASSIVE-SHIELD case study

RFML and Unknown Signals

ML-assisted classification, source-held-out evaluation, and unknown-family detection are treated as engineering workflows that need repeatable data, labels, scoring, and claim boundaries.

SDR prototype development

Experimental Integration Paths

FM, sub-GHz, VLF/LF/MF, and tighter cellular entity fusion are useful integration directions, but they should be labeled experimental unless a project supplies validation data.

View platform overview

What this helps teams do

Connect RF observations to operational outcomes.

Reduce RF triage

Package detections, signal evidence, and operator views so teams can inspect meaningful events faster.

Correlate sensors

Normalize events from SDR, cellular survey, WiFi, BLE, GPS, and defense-oriented prototype cues into shared timelines.

Map RF context

Use route-tagged observations and location-aware evidence to support coverage mapping, source-area estimation, and next collection.

Accelerate evaluation

Use public-safe demos, deterministic replay, and evidence pages to move from concept to partner review.

View Platform Overview Read Case Studies