Fifth-generation (5G) networks are now entering a stable phase in terms of system architecture and commercial release, and the identification of the advanced features that will shape the evolution of 5G into the sixth generation (6G) of mobile network systems has already started. Despite being at early stages of conceptualization, some key aspects of the future infrastructure have been identified by the community: 6G will bring a paradigm shift from connected things to connected intelligence, supporting even more stringent KPI requirements than 5G, and global coverage (air, ground, and underwater). Therefore, there are strong expectations that Artificial Intelligence (AI) will permeate the 6G network infrastructure, allowing for much swifter and more effective decision-making in scheduling, control and orchestration operations of the end-to-end communication systems. Ultimately, this will allow 6G to support ambitious performance targets such as near-zero latency, apparent infinite capacity, or 100% reliability and availability, so as to support new and diverse classes of innovative mobile services.
The BRAIN project will contribute to making this vision of 6G as a network augmented via pervasive artificial intelligence a reality, by addressing the two main roadblocks. On the one hand, existing AI models employed for network management are black boxes, and their complete lack of transparency is a clear barrier for adoption: here, BRAIN aims at proposing new AI tools for network management that are explainable and trustworthy by design and specifications on robustness that allow to benchmark existing AI models. On the other hand, the disruptive softwarization of the network architecture has opened new opportunities for a deep integration of AI into the future 6G infrastructure that have yet to be explored: here, BRAIN will investigate novel approaches for the design, implementation and evaluation of in-band network intelligence, i.e., AI models that run directly into the user-plane programmable switches, operating at line rate over the transiting data traffic, and laying the foundations to a truly AI-native 6G network.