Bascom Hunter Technologies, Inc. — Department of Defense SBIR Phase I: SOCOM211-003

Description

Radio Frequency (RF) Fingerprinting is an important capability for the warfighter to identify threats and protect against malicious attacks on the RF spectrum. Most RF Fingerprinting techniques work by identifying small variations in a transmitted signal that are introduced by hardware components (capacitors, inductors, etc) of the transmitter within the non-linear regions of operation. Neural Networks provide a way to identify these small variations which can act as a “fingerprint” for the specific RF emitter. Photonic Neural Networks provide an efficient, high bandwidth, low latency method of implementing certain Neural Network Topologies. Other topologies, such as Convolutional Neural Networks (CNNs) are better implemented in the electrical domain. A Hybrid (photonic + electrical) solution is proposed for RF Fingerprinting. Achieving stand-off distances for RF Fingerprinting requires the ability to process signals with a low Signal to Noise Ratio (SNR). We propose an RF Noise Cancelling Frontend to our design that could be implemented with a Microwave Photonic Canceller (MPC). We anticipate that this proposed architecture will demonstrate the feasibility to accurately fingerprint signals with 5 dB SNR, at 100-550 us latency and with a 5 GHz Bandwidth. The low SNR capability will enable stand-off distance fingerprinting. The low latency will provide a path to real-time identification. And finally, the high bandwidth will allow multiple channels to be assessed simultaneously without any additional hardware – improving Size, Weight and Power (SWaP).