Nicola Di Cicco

From 15/09/2023to 14/03/2024

Open Ideas

Robust X-Haul Management Based on Uncertainty-aware ML

Columbia University

I am a final-year PhD student in Telecommunications at Politecnico di Milano, Italy, under the supervision of Prof. Massimo Tornatore. Before that, I got my B.Sc. and M.Sc. in Telecommunications Engineering at University of Bologna, Italy. My work is mostly focused the integration of Machine Learning with communications network optimization, management, and control. Specifically, I am researching how to design intelligent agents that can learn to optimize a communication network, and how do they compare with classical optimization techniques. I enjoy skiing, hiking, tennis, and video games.

The purpose of this project is to provide network operators with a reliable Machine Learning-based resource allocation algorithm for X-Haul networks. As X-Haul networks comprise millimeter-wave wireless backhaul, resource allocation must be robust to uncertain events (most prominently, inclement weather conditions) that may disrupt the communication channel. The key challenge lies in ensuring that resource allocation can cope with ever-increasing application demands, e.g., in terms of user throughput, latency, and availability, while adapting to the temporal evolutions of the wireless channel. The outcome of this project will be validated on real telemetry data from a real-world millimeter-wave backhaul network collected over the course of several months, which is an unprecedented data collection effort compared to the current state-of-the-art.

The results of this research project will be submitted for publication in a high-impact conference or journal. The associated software will be also publicly provided to the research community with an open-source license, potentially enabling downstream industrial impact. Finally, I expect the outcome of my expedition to spark a long-lasting collaboration between Politecnico di Milano and Columbia University.

I expect the outcome of this project to be impactful in several orthogonal aspects, namely:

• Joint publication of the main results in a high-impact conference/journal,
• Publication of the research artifacts (e.g., software, and possibly data) associated to the project with a permissive open-source license to facilitate downstream industrial impact,
• Sparking a long-lasting line of cooperation between Politecnico di Milano and Columbia University, potentially leading to other student exchange programs,
• Opening up joint applications to further research funding opportunities,
• Starting a new line of research on X-Haul network management at Politecnico di Milano, resulting in both joint publications with Columbia University and independent projects, involving first-year PhD and final-year Master students,
• Development of innovative teaching material in Next-Generation Internet subjects based
  on remote experimentation on US hardware testbeds and community networks.