This project is a research collaboration between Fondazione Politecnico di Milano (FPM) and SIAE Microelettronica. The project is organized in two main research topics:

1) Optimization of radio networks, a collaboration between FPM and SIAE Microelettronica
2) Optimization of optical networks, a collaboration between FPM and SM-Optics.

This project is a renewal of an initial project between FPM and SIAE Microelettronica, titled “Innovative optimization techniques for the design and management of radio and optical networks”, carried on in the period 2019 and May 2021. The current project is planned for the period 2022 to 2024.

1) Optimization of radio networks, a collaboration between FPM and SIAE Microelettronica

Following recent trends for the management of radio networks, the aim of the project is to make use of large amounts of monitored data, and use such data to improve and automate the management of radio networks. More specifically, we develop new network management tools for the prevention and identification of failures that can be due both to unpredictable events (e.g., weather related) and to natural network degradation (e.g., related to aging of physical components). The underlying optimization and decision-making technique in the project come from recent rising subareas of Machine Learning and Artificial Intelligence, as Deep Learning, Reinforcement Learning, Transfer Learning, Explainability and Online Learning.

2) Optimization of optical networks, a collaboration between FPM and SM-Optics

SM-Optics is an optical-network equipment vendor located in Milan that sells innovative optical-network solutions to support current network deployments for 5G networks and FTTx worldwide. In the project Politecnico provides novel and customized optical-network optimization solutions to assist SM-Optics in the deployment of its products. Politecnico provides modelling of the physical performance of optical networks equipped with particular node architecture, called “filterless”. Starting from an accurate modelling of physical layer characteristics, in our research we develop novel optimization methodologies and algorithms for the design of optical networks considering the aggregation of traffic at several client bit rates (e.g., enabling co-existence of 10G, 100G, and 200G optical circuits) and the requirements for protection of traffic. The final goal is to provide solutions that allow for cost minimization, performance maximization, energy consumption minimization, and maximum network expandability.

The optimization methodologies in the project include Integer Linear Programming (ILP) models, heuristic approaches (including greedy and metaheuristics), and novel rising Machine-Learning-based approaches for combinatorial optimization.