AI@EDGE
A Secure and Reusable Artificial Intelligence Platform for Edge Computing in Beyond 5G Networks
Key facts
- € 8.001.247,50 Budget
- 4 Use Cases
- 19 Partner organisations
- 19 Deliverables
- 8 EU Countries
- 36 months duration
Project Scope
AI@EDGE aimed to develop a connect-compute fabric – specifically leveraging the serverless paradigm – for creating and managing resilient, elastic, and secure end-to-end slices. Such slices were to be capable of supporting a diverse range of AI-enabled applications. Privacy-preserving machine learning and trusted networking techniques were used to ensure each stakeholder could use the platform without disclosing sensitive information.
In AI@EDGE European industries, academics and innovative SMEs committed to achieve an EU-wide impact on industry-relevant aspects of the AI-for-networks and networks-for-AI paradigms in beyond 5G systems. Cooperative perception for vehicular networks, secure, multi-stakeholder AI for IoT, aerial infrastructure inspections, and in-flight entertainment were the use cases targeted to maximise the commercial, societal, and environmental impact.
To achieve its goal, AI@EDGE targetted significant breakthroughs in two fields:
- general-purpose frameworks for closed-loop network automation capable of supporting flexible and programmable pipelines for the creation, utilization, and adaptation of the secure, reusable, and trustworthy AI/ML models; and
- converged connect-compute platform for creating and managing resilient, elastic, and secure end-to-end slices capable of supporting a diverse range of AI-enabled network applications.
The AI@EDGE project focused on six main breakthroughs:
- AI/ML for closed loop automation;
- Privacy preserving, machine learning for multi-stakeholder environments;
- Distributed and decentralized connect-compute platform;
- Provisioning of AI-enabled applications;
- Hardware-accelerated serverless platform for AI/ML;
- Cross-layer, multi-connectivity and disaggregated radio access.
Finally, the AI@EDGE platform was validated using four well-chosen use cases with specific requirements that couldn’t be satisfied by current 5G networks according to the 3GPP Rel15 and 3GPP R16 standards, in particular in terms of support for latency-sensitive and highly dynamic AI-enabled applications.