DARPA looks to machine learning for RF defense program

BAE Systems has booked a $9.2 million Defense Advanced Research Projects Agency contract to develop new machine learning algorithms and other related tools for users to better identify radio frequency signals.

Phase one of the Radio Frequency Machine Learning System program will see BAE focus its creation work on using cognitive approaches to make the algorithms with feature learning techniques.

Scientists will also work to create algorithms that help users determine signals that are important versus those that are not in real time through a deep learning approach, BAE said Tuesday.

DARPA envisions this program as further advancing technology for autonomous vehicles and image and speech recognition through data-driven machine learning research. The agency particularly wants to counter adversaries’ efforts to disrupt RF signals and examine how machine learning can help in that.

Internet-connected devices such as phones, drones and sensors are another factor to consider in RF spectrum usage, according to BAE. Those devices can be used to hack, spoof and disrupt the RF spectrum.