Collaborative Research: VINES: Track 1: ARMANI - AI-native RAN MANageability and Intelligence for Next Generation Networks

Next-generation wireless networks, such as 5G/6G, require smarter and automated tools to manage the increasing network complexity. This collaborative project brings together investigators from Yale University and North Carolina State University to develop ARMANI (Artificial Intelligence-native Radio Access Network Manageability and Intelligence for Next Generation Networks), a unified framework that embeds artificial intelligence directly into the core of how wireless networks are monitored, managed, and controlled. Built on the Open Radio Access Network (O-RAN) architecture, ARMANI enables wireless networks to make faster, smarter, and more reliable decisions automatically, helping connect more people with better service. The ARMANI framework addresses three critical limitations in current O-RAN network management. First, it introduces a shared learning substrate within the RAN Intelligent Controller (RIC), allowing modular network applications, known as xApps and rApps, to exchange learned representations, coordinate decisions, and resolve conflicts efficiently. Second, it designs orchestration mechanisms that dynamically co-locate artificial intelligence workloads alongside Radio Access Network functions, harnessing idle compute capacity through fine-grained observability and predictive scaling. Third, it develops a foundation model-driven reasoning layer that fuses multimodal telemetry streams, including Key Performance Measurements and device logs, to deliver proactive, context-aware control through novel multi-modal AI alignment of Large Language Model (LLM) and time series encoders. ARMANI has the potential to significantly improve the efficiency, reliability, and adaptability of next-generation wireless networks, benefiting critical applications such as public safety communications, autonomous systems, and immersive media. By improving resource efficiency and network resilience, the project supports equitable access to advanced connectivity in all environments. The project will contribute open-source tools, datasets, and experimental methodologies to a wide research community, fostering wider adoption of artificial intelligence-native network control. Education and workforce development are integral to the project, with graduate and undergraduate students trained at the intersection of wireless networking, machine learning, O-RAN, and systems engineering. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.