As AI moves closer to the source of data – on drones, sensors, or mobile ground units – there is a growing need for reliable, low-latency intelligence in disconnected, high-risk environments. However, deploying machine learning in such scenarios presents significant challenges: limited connectivity, low bandwidth, and the critical requirement for a mobile hardware that can securely run AI workloads at the edge.

Oracle Roving Edge Device (RED) for Federated Learning at the Edge

To tackle these challenges, Scaleout partnered with Oracle to integrate their Edge AI platform (Scaleout Edge) with Oracle Roving Edge Device (RED). The 2nd generation RED device equipped with GPUs allowed Scaleout to train AI models locally on a ruggedized, portable hardware platform that delivers unparalleled processing power, seamless connectivity, enhanced GPU performance and ironclad security even in disconnected locations. According to Salman Toor, CTO of Scaleout systems:

Bringing AI to the edge isn’t just about performance, it’s about trust, security, and being able to adapt in tough conditions. By combining our federated learning platform with Oracle RED, we’ve built something that’s ready for the field and built for real-world defense use— Salman Toor, CTO, Scaleout Systems

Field Proven Solution

This combined Scaleout Edge and RED platform was put to test in the FEDAIR project which was part of NATO’s DIANA Challenge Programme. FEDAIR demonstrated how drones and ground units collaborate using Federated Learning and enable continuous learning at the edge with Oracle RED playing a pivotal role as the hardware backbone. Local command centers, powered by RED, acted as aggregators, enabling synchronization even when disconnected from central command. This enhanced resilience and operational agility, showcasing how Oracle RED enables resilient federated machine learning deployments with secure local storage for enhanced information security under real-world constraints of no raw data transmission and operation without constant network access.

Continuous Edge-Based Machine Learning
Continuous training and coordination across edge devices, aggregators, and cloud, without centralizing data

 

Oracle’s Roving Edge Device (RED) and Scaleout’s Edge AI platform: An Integrated Edge AI Stack

The integrated Edge AI stack comprising of Scaleout’s Edge platform and Oracle’s RED was tested in three core deployment models based on Scaleout’s federated learning framework architecture:

  1. RED as a Scaleout Edge Unit: Acting as a high-performance client node, RED performs local training on sensitive data and contributes updates to the wider federation. The massive compute power of RED supports demanding workloads, including computer vision and large language models.
  2. RED as a Scaleout Edge Combiner: In more complex deployments, RED serves as an intermediary, aggregating model updates from dispersed edge devices like drones or vehicles and handling asynchronous communication. This supports scalable training across mobile and distributed networks.
  3. RED as a Scaleout Edge Framework Node to run full Scaleout Edge stack: In fully disconnected scenarios, RED can run the entire federated learning stack independently, enabling localized model training among multiple edge devices without requiring a persistent cloud connection.

Three deployment models for RED across ScaleOut Edge Architecture

The Integrated Edge AI stack shows how RED with GPUs supports flexible federated learning at the edge, cloud-connected, partially connected, or fully disconnected. In each of these roles, integration with Oracle Cloud Infrastructure (OCI) and Oracle’s Sovereign Cloud services enables secure data handling and continuity of operations when reconnected.

We have deployed Scaleout Edge AI platform on Oracle RED in multiple roles — from edge client to full federated node — supporting secure AI training even in disconnected environments. Its flexibility and compute power make real-world edge deployments faster and more reliable. — Stefan Hellander, CISO, Scaleout Systems 

Edge AI, Ready for Deployment

For defense organizations that require a robust cloud-integrated service built to run mission critical, time-sensitive applications at the edge—whether connected or disconnected—Oracle Roving Edge Infrastructure serves as the linchpin. When combined with Scaleout’s Edge platform, it delivers a mission-ready, secure edge AI solution built for mission success. Discover how Oracle Roving Edge Infrastructure and Scaleout Edge can transform your operational capabilities today.