Volumetric Automatic Radiofrequency Inference of Urban Sources (VARIOUS)

Exponential growth in radiofrequency (RF) emitters and spectral density across all domains in unpopulated and urban centers creates an intractable emitter geolocation challenge demanding highly innovative solutions. Communications between units, command orders, drone control, remote sensors, and radar to identify, track, and target adversaries depend on RF emitters. Effectively targeting RF emitters enables tactical and operational advantages that limit adversary ability to understand and respond to changing battlefield conditions. However, it is infeasible to calculate the paths of radio waves in complex urban and suburban environments in operationally relevant time frames on a processor light enough to deploy to a battlefield (e.g., on uncrewed aerial vehicles (UAVs)). The complex effects of objects on radio waves are hard for humans to reason about or visualize, especially for multipath signal propagation. Volumetric Automatic Radiofrequency Inference of Urban Sources (VARIOUS) is an artificial intelligence (AI)¢based tactical capability to identify and localize RF sources from a UAV and map them in a collaborative augmented/virtual reality environment for analysis and planning. VARIOUS uses 3D maps of environments to generate machine learning source localization and signal-type identification models for deployment onboard a commercial off-the-shelf processor using explainable AI techniques to characterize the resulting probability metrics.