Artificial intelligence can be a game-changer in Intelligence, Surveillance, and Reconnaissance (ISR), but requires significant R&D, rigorous testing, and governance for field readiness. As the range of new threats has broadened, ISR technologies have kept pace, advancing both hardware and software to support joint ISR goals. AI is advancing rapidly to support today’s biggest challenges in ISR:
- Radar Warning Receivers and ELINT collection systems must be pre-programmed before flight to Detect, Identify, Locate, and Report (DILR) KNOWN threats within the AOR.
- New unknown threats, war reserve modes, and frequency/modulation/etc. Agile signals place the warfighter at risk.
- Intelligence analysis, Specific Emitter Identification (SEI), and Battle Damage Assessment (BDA) of RF threats before and after mission execution are time-consuming and impractical during high op-tempo.
- New commercial technologies, specifically 5G, utilize advanced beamforming, frequency diversity, and UHF/VHF-to-mmW frequencies often found in military systems today, making it very difficult to detect DILR-specific threats.
- Cybersecurity threats within the RF chain are difficult to identify without significant transmission overhead (encryption, handshaking, etc.)
Leonardo Electronics US is a leading innovator in aerospace, defense, and security, and provides the United States Department of Defense (DoD) and its prime contractors with ISR systems, such as radar, infrared systems, electronic warfare, and avionic systems, with customization and software integration support by its Huntsville, AL facility. Leonardo has invested significant R&D and testing efforts into advancing its AI/ML integrated capabilities with applications to radar, IR, and electronic warfare technologies.
Initial Radio Frequency Machine Learning (RFML) capability demonstrated the power of AI for emitter identification using only signal data. Combining multiple sensor streams creates a truly robust “fingerprint” to characterize emitters.
Leonardo has now integrated additional platforms to ingest data from:
- LEO satellites which provide wide-area intercepts of RF signals. This adds geographical context and cross-cueing between the aerial and space domains.
- Native radar and EO/IR sensors like Leonardo’s Osprey AESA radars and airborne multi-spectral targeting systems. These provide high-resolution tracking and visuals for enhanced geolocation and characterization.
- Diverse intelligence sources via modular open architectures, which can augment the RF data with images, cyber fingerprints, and even free-text data mining of open-source intelligence.
Image 1, below, demonstrates the process flow for Leonardo's novel approach to RFML in signal processing, which can both classify and identify a potential threat.
Image 1: Novel AI/ML approach for ISR that combines classification with identification, with >90% confidence in identification.
Multi-modal neural networks can combine these heterogeneous data types, with attention mechanisms to learn which features are most relevant for characterization tasks. This provides a diverse evidentiary base for identification.
The global sensor network enables persistent monitoring and rapid cueing to focus multiple assets on emitters of interest. By combining space, air, and ground collection, Leonardo can provide the spectrum coverage and resolution needed for comprehensive Emitter ID.
By leveraging diverse intelligence, not just technical parameters, we characterize the operational context around transmitters. This paints a human picture of who, what, when, where, and why regarding RF emissions and establishes patterns of life that can be an early warning when the out-of-the-ordinary occurs.
Open architecture ingests new data sources quickly, allowing warfighters to capitalize on emerging commercial capabilities. The AI models automatically adapt, providing ever-increasing fidelity to gain information advantage across the multi-domain battlespace.
This expanded sensor fusion represents the next evolution of Leonardo’s emitter ID capabilities. AI is the glue that binds disparate data into integral situational awareness. By synthesizing multiple perspectives, joint ISR can gain an unmatched view of the electromagnetic and operational environment.