Project logo

bottone focused_new

ADAPTO aims at defining a management framework for 5G and beyond network that allows infrastructure providers to properly manage the computational resources required to accommodate the dynamic instantiation of network functions by combining QoS requirements with at-scale energy saving objectives. This goal will be pursued by integrating both model-based and AI-based optimization techniques to NFV-enabled software-defined networks.

ADAPTO is part of Spoke 8 – Intelligent and Autonomous Systems 

Project PI: Giorgio Ventre

At M9 the ADAPTO project has established a preliminary investigation of resource monitoring and data collection techniques for cloud-native Telco infrastructures in the context of 5G and beyond communication networks to identify algorithms and techniques to be implemented in the ADAPTO framework.

At M18 the ADAPTO project has concluded a preliminary design of the ADAPTO framework, by sketching the overall architecture and identifying the purpose of its components. Deliverable D3.1 describes the framework and presents the methodological approaches, both model-based and data-driven, that are pursued in the framework components to facilitate intelligent scaling of network functions, edge-to-cloud transitions, and optimal selection of 5G backhauls. 
The emergence of the Multi-Access Edge Computing paradigm requires the ability for Telco operators to manage computational and communication resources scattered in a distributed infrastructure.

At M18, ADAPTO has carried out an overall design of a framework that orchestrates resources to be allocated by telco operators to instantiate both network functions and third-party services. The framework relies on collaboration among distributed agents located both at the edges and in centralized datacenters.

Scientific outcomes:

  • 5 papers published at international conferences or journals;
  • 1 paper currently under revision for a journal.


Papers:
Laura Carnevali, Marco Paolieri, Benedetta Picano, Riccardo Reali, Leonardo Scommegna, and Enrico Vicario. A quantitative approach to coordinated scaling of resources in complex cloud computing workflows. 19th European Performance Engineering Workshop (EPEW 2023), 2023 - https://doi.org/10.1007/978-3-031-43185-2_21 

Alessio Botta, Roberto Canonico, Annalisa Navarro, Giovanni Stanco, Giorgio Ventre. Adaptive Overlay Selection at the SD-WAN Edges: A Reinforcement Learning Approach with Networked Agents. Computer Networks (COMNET), special issue on Network Softwarization and Intelligence at the Edge, Volume 243, Article 110310, April 2024 - https://doi.org/10.1016/j.comnet.2024.110310 

Laura Carnevali, Marco Paolieri, Riccardo Reali, Enrico Vicario. Compositional Safe Approximation of Response Time Probability Density Function of Complex Workflows. ACM Transactions on Modeling and Computer Simulation, 2023 - https://doi.org/10.1145/3591205 
Industrial partners involved in the project are also active in standardization processes and contribute in keeping the project outcomes aligned with the market expectations.
Program partners: Cascade call partners:
ADAPTO contributes to the RESTART efforts targeted at Grand Challenge "Make the network a platform for programming and running applications" by investigating how traditional model-based approaches can be combined with Machine-Learning based techniques for resource management at scale in distributed infrastructures aggregating centralized clouds and small-scale edge datacenters that provide computational and data processing capabilities to support time-sensitive services.

  • Paper "Towards a Highly-Available SD-WAN: Rapid Failover based on BFD Protocol" has been presented by Giovanni Stanco (UNINA) at the IEEE Conference on Network Function Virtualization and Software Defined Networks (IEEE SDN-NFV) 2023 conference in Dresden on November 9

Researchers involved: 96

Collaboration proposals:
ADAPTO welcomes research collaborations with projects/researchers interested to the definition of APIs to export network function metrics and to provide application requirements to the network (a la GSMA Open Gateway) in next generation networks.

For any proposal of collaboration within the project please contact the project PI.