We propose an innovative system to monitor RF spectrum and rapidly process High Sample Rate IQ data in real or near-real time for emitter detection and identification. We base this solution on a compressive sampling front end to generate IQ data with much lower data requirements than a standard Nyquist process coupled with machine learning algorithms designed to be used in lieu of or to augment the standard method of PDW generation and classification. This unique combination of technologies results in a system that offers full compliance with the requirements.
Benefit: Our solution can rapidly process the IQ data before PDW formation to exploit unused information while lowering the computational requirements and data-throughput. We provide complete RF spectral monitoring and emitter detection over extended bandwidths of interest and operate under real-time constraints on existing hardware platforms. By operating at the sensor level and sampling intelligently, these subsystems will reduce bandwidth constraints to downstream processing while maintaining complete coverage of the RF environment.
Keywords: cognitive sampling, cognitive sampling, convolutional neural networks, Machine Learning, Electronic Warfare, cognitive surveillance
