Traffic Collection, Contextual Analysis, Data-driven Optimization for 5G – CoCo5G

Since 2021, 5G is being deployed worldwide. Its coverage is slowly progressing and is expected to accelerate in the incoming years. Emerging trends and interest toward new and unknown 5G and MEC applications might be observable in aggregated traffic profiles of the preliminary 5G deployments, in particular in regions expected to undergo an anticipated deployment, such as the Ile-de-France region for the 2024 Olympic Games.

In this project, we aim at precisely working on the analysis of 5G traffic aggregates for the period from 2023 through 2025, in order to anticipate and better motivate 5G deployment and configuration strategies and also anticipate 6G evolution. By leveraging on the skill-set acquired in previous ANR projects, i.e., the ABCD and CANCAN projects, the consortium brings together a major telecom and IT integrator (Thales), a major mobile access service provider (Orange), and three academic labs: the Cnam CEDRIC lab with its Networks and IoT Systems group, the Inria Agora group, and Imdea.

The key objectives of the project are: (i) to collect first-of-its-kind longitudinal dataset of 4G+5G data traffic, (ii) to understand how 5G will be changing usages over time and space, and (iii) to provide an early study informed by large-scale real-world measurements of means to foster the performance, adoption, and evolution of 5G.

We plan to complement aggregate data extraction tools employed for 3G/4G mobile data traffic with new methods for 5G data traffic collection. We target per-service data collection across all technologies as well as extracting IoT data. The collected data and the insights gained on service usage in 5G represent the ideal ground to test machine learning models and optimization algorithms, and apply them to practical use cases, including 5G sustainability in terms of energy consumption, low-latency edge computing services with a particular focus on vCDNs, and special events detection.