BAE Systems won a contract worth up to $4.7 million from the Defense Advanced Research Projects Agency (DARPA) to add machine learning to systems used in electronic warfare. The company plans to put its Chimera technology in so-called signal intelligence systems to make sense of radio frequency signals intercepted from adversaries and anticipate threats.
Chimera—which stands for controllable hardware integration for machine learning-enabled real-time adaptivity—is a reconfigurable hardware platform. Developers can use it to detect and decode radio frequency signals sent out by communications or radar systems. The system is necessary for sifiting through increasingly crowded RF spectrum on the battlefield, the company said.
"Chimera brings the flexibility of a software solution to hardware," Dave Logan, general manager of the defense contractor's C4ISR systems business, said in a statement. Communications, radar and other electronic warfare systems could also feature the hardware platform in the future. Chimera integrates a reconfigurable array, front-end, full transceiver as well as digital pre-processing.
"Machine learning is on the verge of revolutionizing signals intelligence technology, just as it has in other industries,” Logan added. The contract is BAE's second under DARPA’s Radio Frequency Machine Learning Systems, or RFMLS, program launched in August 2017. The first focused on developing machine learning algorithms able to discriminate a wide range of RF signals.
Chimera is capable of adapting its RF configuration in real time, improving performance over previous types of hardware. The system supports on-the-fly performance trade-offs to boost sensitivity, selectivity and scalability. The open architecture also allows anyone to develop algorithms for Chimera, which will make it easier to upgrade and less likely to become obsolete, BAE said.