Architecture

The Strangeloop
technology stack.

From raw RF to actionable intelligence — entirely on the sensor, entirely in real time. A purpose-built stack for edge AI inference in denied, degraded, and disconnected environments.

Signal Processing Pipeline
Layer 1
RF Front-End Wideband ADC

DC–6 GHz
6.144 GSps
16-bit resolution

Layer 2
DSP FPGA Pre-Processing

Channelization
DDC / DUC
FFT engine

Layer 3
AI Inference Edge Neural Engine

INT8 quantized
<2 ms latency
10k+ signal types

Layer 4
Fusion TDOA Geolocation

Multi-node mesh
Sub-meter accuracy
Real-time tracking

Layer 5
Output Operator Interface

REST / gRPC API
TAK integration
ATAK / WinTAK

Total pipeline latency from antenna to classified alert: <2ms on SWaP-optimized hardware. Zero data leaves the node unless explicitly commanded.

Edge AI

Neural inference
at the antenna.

Strangeloop's inference engine is built on a custom neural architecture trained on the world's most comprehensive RF signal dataset — over 4 terabytes of annotated captures spanning commercial, government, and adversarial emitters. Models are quantized to INT8 and run entirely on-device. No GPU cluster, no cloud endpoint, no network dependency.

Model Architecture

Temporal convolutional networks (TCN) with attention mechanisms. Trained on IQ samples, not demodulated bits.

Continual Learning

On-device few-shot adaptation to novel waveforms. New signal profiles deploy via signed OTA update.

Uncertainty Quantification

Every classification carries a calibrated confidence score. Low-confidence detections trigger expanded collection.

Adversarial Hardening

Models trained with adversarial examples to resist intentional waveform manipulation and evasion.

Operating temp
−40°C to +85°C
MIL-STD
810H, 461G, 1275E
Weight
<2.1 kg
Power
18 W nominal
Interfaces
10GbE, USB-C, SMA
Geolocation Engine

TDOA

Time-difference-of-arrival across a mesh of Strangeloop nodes. Nanosecond-precision timestamping with hardware-level PPS sync.

AoA

Angle-of-arrival via phased-array antenna processing. Single-node bearing fix in under 500 μs. Multi-node triangulation for 3D position.

Hybrid Fusion

Kalman-filter fusion of TDOA, AoA, and RSSI measurements. Adapts dynamically to node availability and SNR conditions. Sub-meter CEP50 at 2 km.

Accuracy vs. Range
Range CEP50 Nodes Required
500 m<0.3 m3
2 km<0.8 m3
10 km<5 m4
50 km<50 m4+
Timing & Sync
PPS sourceGPS / GNSS or disciplined oscillator
TDOA resolution1 ns
Sync error budget<5 ns RMS
GPS-denied fallbackDisciplined TCXO, <50 ns/hr drift
Adaptive Spectrum Monitoring

Autonomous. Adaptive.
Always-on.

Strangeloop's monitoring engine doesn't require an operator to specify what to look for. It continuously maintains a spectral baseline, detects deviations, and autonomously adjusts scan parameters to track threats.

Priority queuing ensures the most operationally relevant signals receive maximum dwell time — without dropping awareness of the broader electromagnetic environment.

  • Dynamic frequency hop tracking — follows FHSS signals across the band
  • Automatic baseline calibration on deployment — no manual survey required
  • Anomaly detection flags new emitters entering the environment in <100ms
  • Operator-defined priority bands receive dedicated tuner allocation
Frequency Coverage
HF / VHF3–300 MHz
UHF300 MHz–3 GHz
SHF (partial)3–6 GHz

Extended frequency ranges available via optional front-end modules.

Scan Performance
Instantaneous BW6.144 MHz – 6 GHz (multi-tuner)
Scan rate (full band)<10 ms per sweep
Minimum dwell1 µs per channel
Concurrent monitorsUnlimited (software-defined)
Learn more

Ready to see it in context?

View Use Cases Request Technical Briefing