Where Are The (Cellular) Data?

International audienceNew generations of cellular networks are data oriented, targeting the integration of machine learning (ML) and artificial intelligence solutions. Data availability, required to train and compare ML-based networking solutions, is therefore becoming an important topic and a significant concern. Operators do collect data, but they rarely share it because of privacy concerns. This paper starts by reviewing the few publicly available cellular datasets, which created bursts of innovation with their release. The scarcity of such data is so acute that researchers are collecting network data using their own tools, developed in-house and covered by the second part of this survey