Scaleout has been selected to participate in NATO’s DIANA Challenge Programme with the FEDAIR (Federated Aerial Intelligence for Recon) project. This recognition highlights the growing importance of innovative solutions for defense challenges. At its core, FEDAIR addresses a pressing problem: how to keep machine learning (ML) models effective in fast-changing environments like conflict zones, where adapting to new information is critical.
ML models are not static tools—they rely on up-to-date data to maintain their accuracy. Over time, as situations evolve, models trained on older data become less effective. In conflict zones, this degradation happens quickly. New threats, altered landscapes, and shifting conditions demand models that can adapt in real time. Without updates, critical systems relying on ML risk losing their relevance when they are needed most.
However, traditional methods of updating ML models rely on centralized data collection and processing. In environments where communication networks are unreliable or where data sensitivity is a concern, this approach breaks down. Centralizing data for training becomes slow, risky, or outright impossible.
FEDAIR uses federated learning (FL), a decentralized method that eliminates the need for centralized data. Instead, ML models are updated directly on the edge—such as drones, operator tablets, or field command posts—using data collected locally. Only the model updates are shared with a central server, not the raw data itself. This drastically reduces communication needs while protecting sensitive information. By addressing these issues, FEDAIR aligns with NATO’s DIANA information security challenge, which prioritizes secure, decentralized data handling and resilience in multi-domain environments.
By allowing models to adapt at the edge, FEDAIR ensures they remain effective in responding to new conditions. It builds on existing successes of federated learning in industries like healthcare and IoT, adapting the technology to meet the unique challenges of defense operations.
%25201.webp)
Conflict zones amplify the challenges of maintaining effective ML models. Communication networks are often contested or disrupted, data security is paramount, and the environment changes unpredictably. FEDAIR addresses these realities.
For example, field operators equipped with FEDAIR technology could analyze and learn from newly captured data without transmitting sensitive information. This capability allows them to detect new threats or adapt to evolving conditions without depending on a stable network connection. It also enhances operational security, as sensitive raw data never leaves the local area.
Another critical advantage is resilience. FEDAIR’s decentralized approach ensures operations can continue even when connectivity is intermittent. Each edge device operates independently, maintaining adaptability and effectiveness regardless of external disruptions.
FEDAIR represents a practical step forward in ensuring ML systems can adapt to complex and high-risk environments. Its ability to update models securely and efficiently at the edge makes it an invaluable tool for improving situational awareness and decision-making in conflict zones.
As Scaleout takes part in the DIANA Challenge Programme, FEDAIR highlights the importance of developing technologies that address real-world limitations while building resilience and adaptability into critical systems. It’s an approach designed not just for today’s challenges, but for the evolving demands of tomorrow.
Supported by
.png)
About NATO DIANA
DIANA is the Defence Innovation Accelerator for the North Atlantic, a NATO organisation with a mission to locate and accelerate dual-use innovation across the Alliance. DIANA provides technology developers with the resources, networks and expertise to address critical defence and security challenges, to create a more peaceful and resilient future.
Learn more in this announcement from NATO DIANA on LinkedIn:
https://www.linkedin.com/posts/nato-diana_diana-announces-2025-innovator-cohort-activity-7272565935821991936-hgb4/
And here on their web page:
https://www.diana.nato.int/about-diana/2025-cohort-of-companies.html
%201.svg)
