Emilio Paolini

From 04/12/2024 to 30/04/2025

Open Ideas

ADAPTIVE EDGE-AI DEPLOYMENT IN NEXTG WIRELESS NETWORKS

Saint Louis University

Emilio Paolini is an Assistant Professor at Scuola Superiore Sant’Anna, specializing in AI-driven solutions for next-generation wireless networks. His research focuses on neuromorphic computing, real-time network intelligence, and AI optimization in constrained environments. He completed his Ph.D. in Emerging Digital Technologies at Scuola Superiore Sant’Anna, working on AI methodologies for wireless networks and neuromorphic hardware. He received the NGI Enrichers Postdoc Fellowship, and won Best Paper Award at the 2024 IEEE 25th International Conference on High Performance Switching and Routing. His work advances AI-powered network resilience, efficiency, and sustainability, contributing to the future of intelligent communication systems.

The increasing complexity of NextG wireless networks demands intelligent, adaptive AI solutions to optimize performance, reduce latency, and improve energy efficiency. However, the integration of Deep Learning (DL) into these networks has led to rising computational costs and inefficiencies, particularly due to the diverse and evolving nature of base station hardware. Existing approaches lack the flexibility to dynamically adjust AI models in real time, limiting their effectiveness in changing network conditions.

This project aims to develop an adaptive AI framework capable of optimizing edge AI deployments in NextG networks. By leveraging advanced learning techniques and real-time model adaptation, it ensures efficient resource utilization while maintaining high performance. The solution will be validated through a joint testbed between Scuola Superiore Sant’Anna and Saint Louis University, contributing to research publications, open-source development, and future advancements in sustainable and intelligent wireless networks.

The expected results of this project include the development of an adaptive AI framework capable of optimizing model complexity in real time based on network conditions and hardware constraints. By leveraging energy-efficient AI deployment strategies, including hardware-aware compression techniques and dynamic model adaptation, the project aims to minimize latency and computational overhead in NextG wireless networks. Additionally, the project aims at enhancing the scalability and sustainability of AI at the edge by integrating federated learning strategies that improve distributed model training efficiency while reducing communication costs. The validation of these approaches will be conducted through a joint testbed between Scuola Superiore Sant’Anna and Saint Louis University, demonstrating their feasibility in real-world network environments.

The NGI Enrichers fellowship is expected to have a strong impact on both scientific progress and international collaboration in AI-driven wireless networks. By developing and testing adaptive AI models for efficient network intelligence and federated learning, this project contributes to the advancement of innovative technologies in NextG networks. The integration of real-time model adaptation, congestion control, and in-network aggregation at the gNB introduces novel approaches for improving energy efficiency and reducing communication overhead, reinforcing the project’s technological innovation and scientific validation. A key outcome of the fellowship is the deployment of a fully operational testbed at Saint Louis University, enabling experimental validation of AI-driven networking strategies in real-world conditions.

Overall, this experience has been invaluable for my professional growth, expanding my research perspective and strengthening my expertise in AI-driven wireless networking. The skills, connections, and insights gained through this fellowship will support my future work and open opportunities for further contributions in research and technology development.