Luca Fattarelli
A Secure Federated Reinforcement Learning Approach for Traffic Engineering.
Rel. Alessio Sacco, Guido Marchetto, Doriana Monaco. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2026
|
Preview |
PDF (Tesi_di_laurea)
- Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (4MB) | Preview |
Abstract
The growing adoption of Machine Learning techniques for traffic engineering has enabled the development of routing systems capable of adapting to dynamic network conditions. Among these approaches, Reinforcement Learning has emerged as a promising solution for learning routing policies directly from interactions with the network environment. However, training such models often requires access to sensitive information, including network topologies, traffic demand matrices, and routing policies. As a result, collaboration across multiple network domains remains challenging, since operators are generally unwilling to share data that may reveal details about the internal operation of their infrastructures. To address this challenge, this thesis investigates a privacy-preserving Federated Reinforcement Learning framework for distributed routing optimization.
The proposed solution enables multiple network domains to collaboratively train a shared routing policy while keeping local network information within administrative boundaries
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Corso di laurea
Classe di laurea
URI
![]() |
Modifica (riservato agli operatori) |
