Revolutionizing SIGINT with Automated Machine Learning


Authored by David Ramirez at General Dynamics Mission Systems featuring Curtiss-Wright Defense Solutions

Operating in the electromagnetic spectrum keeps getting more demanding. In foreign countries, tactical communications might once have been the only signal on the airwaves. Today, however, civilian cell service, Wi-Fi, satellites, and radio waves penetrate every corner of the globe. The majority of these devices operate on expected frequency bands, but many do not. Especially in developing markets with minimal regulation, the prevalence of interfering transmitters is growing rapidly.

These new signals complicate military operations for friendly communications planning and enemy signals intelligence (SIGINT) collection. An allied communication plan must account for existing civilian communications patterns. Finding these patterns can consume vast amounts of time for the few available highly trained military personnel. Identifying and separating civilian from enemy communications is another huge challenge. In many cases a huge backlog of data is available for exploitation, but opportunities are missed due to a lack of resources. A missed opportunity could result in degraded or dropped communications at a critical time. In hostile environments, it could mean a nearby enemy transmission goes unnoticed; missing the precursor to an attack.