Project
Radiant Expanse
Heterogeneous Distributed Digital Signal Processing Library for RF Interference Mitigation
Prime contractor — direct SBIR Phase II award
Design, build, and test of a first-article prototype: a modular, low-SWaP, GPU-accelerated distributed signal processing system for interference suppression, RF fingerprinting, and machine-learning modulation classification, built on software-defined radio hardware with NVIDIA GPU compute. Work included proprietary interference-suppression algorithms, an operator console with live spectrum display and real-time signal classification and recording, a full interface control document, and custom mechanical and enclosure design.
Outcomes
- —Demonstrated recovery of a signal of interest masked by stronger co-channel interference — a critical capability for operations in contested RF environments
- —Real-time RF fingerprinting distinguishing five modulation schemes (BPSK, QPSK, 8-PSK, 16-QAM, 64-QAM)
- —Completed in-house laboratory testing and initial static over-the-air field testing
- —Dual 80 MHz channels across 10 MHz–6 GHz; up to 80 MSPS per RF channel; 10 Gbps internal data path
SBIR Phase IIRF Interference MitigationDSPGPU ComputingDistributed SystemsSIGINT
Project
Dream Catcher
RF (Wi-Fi / Bluetooth / BLE) CVE Threat Detection for Common Vulnerabilities and Exposures
Subcontractor — government subcontract, 2026
Non-recurring engineering to develop a real-time RF (Wi-Fi / Bluetooth Classic / BLE) threat detection capability — a wireless defensive perimeter around fleet vehicles that detects rogue devices, impersonation attempts, insecure connections, and wireless denial-of-service activity. Work spanned algorithm development, device customization, data capture, machine learning model training, and system integration on commercial SDR hardware with NVIDIA edge GPU compute. Our software-defined GPU approach departs from conventional FPGA/ASIC implementations, executing statistical and DSP algorithms as real-time CUDA pipelines.
Outcomes
- —Built GPU receive pipelines for 802.11a/g/n Wi-Fi (full OFDM chain: packet detection, synchronization, channel estimation, multi-rate demodulation, Viterbi decoding, MAC-layer parsing), BLE GFSK advertising, and 79-channel FHSS Bluetooth Classic
- —Developed and validated detection of known CVE attack families across Wi-Fi and Bluetooth — including KRACK, Evil Twin, and BrakTooth-class attacks — verified by comprehensive regression testing
- —Demonstrated live over-the-air operation: sustained Wi-Fi frame detection and a 24-hour Bluetooth Classic capture producing validated packet detections
- —Delivered source code, binaries, a user manual, and a final report as contract deliverables
Wi-FiBluetoothBLECVE DetectionFleet ProtectionThreat DetectionCUDA