This I³ project targets the application of Machine Learning (ML) in aeronautical communications. In contrary to many engineering applications, which view the learning process as a black box, this project puts a particular focus on the understanding and interpretation of the learning process. In short, the target can be described as a fundamental understanding of what ML can and does do in the context of communication networks.

By combining the expertise from our institute with that of the Institute of Communications, ML is considered on the data link layer. In the presence of coexisting legacy systems, opportunistic spectrum access is studied, which is achieved through the learning of the legacy system's behavior, so that future behavior becomes predictable, and the new system can opportunistically access the spectrum when the legacy system does not.