According to Silicon Republic, researchers from Trinity College Dublin and the Adapt Research Ireland Centre for AI-Driven Digital Content Technology are leading a newly funded €2.5 million project called Naira (Native AI for Energy Efficient and Sustainable Radio Access Networks). The three-year initiative, funded through Call 7 of the Disruptive Technologies Innovation Fund, aims to embed artificial intelligence directly within radio access network architecture, enabling what they call “agentic AI” to make local and collective decisions across the network. The project will be coordinated by Professor Marco Ruffini from Trinity’s School of Computer Science and Statistics and ADAPT, along with co-PI Dr. Merim Dzaferagic, with industry partners including Dell Technologies, Red Hat, Intel Research and Development Ireland, Software Research Systems, and Tyndall National Institute. This initiative represents a significant step toward creating networks that can autonomously reconfigure and optimize themselves in real-time, addressing the growing energy consumption challenges in communication infrastructure.
Table of Contents
The Energy Crisis Driving Innovation
What Professor Ruffini’s comments highlight—but don’t fully quantify—is the staggering scale of the energy problem facing mobile networks. Current 5G networks consume approximately 70% more energy than their 4G predecessors despite being significantly more efficient per bit transmitted. The paradox stems from exploding data demand and network density requirements. As we move toward 6G and ubiquitous connectivity, traditional optimization approaches simply won’t scale. The Naira project’s focus on embedding intelligence directly into the RAN represents a fundamental architectural shift rather than incremental improvement. This approach acknowledges that centralized cloud-based AI solutions introduce too much latency for real-time network optimization decisions that need to happen in milliseconds rather than seconds.
The Technical Breakthrough Behind Agentic AI
The term “agentic AI” used in the project description points toward a distributed intelligence architecture where multiple AI agents collaborate across network nodes. This differs significantly from current AI implementations in telecommunications, which typically involve centralized machine learning models analyzing network data after the fact. True agentic systems would enable individual base stations to make autonomous decisions about power management, frequency allocation, and traffic routing while coordinating with neighboring nodes to optimize overall network performance. The technical challenge lies in developing AI models that can operate within the severe computational constraints of edge devices while maintaining the reliability required for critical infrastructure. This requires breakthroughs in lightweight neural networks, federated learning techniques, and real-time inference optimization that the Naira team will need to pioneer.
Industry Implications and Competitive Landscape
The involvement of major industry players like Dell, Red Hat, and Intel suggests this isn’t merely academic research but has serious commercial potential. Intel’s participation is particularly telling, as they’re positioning themselves for the edge computing revolution that 6G will demand. Meanwhile, the parallel investment by Nvidia in Nokia’s AI-RAN initiatives indicates we’re witnessing a broader industry realignment. The competition isn’t just about whose AI technology wins, but which architectural approach will dominate: Naira’s distributed agentic model versus more centralized solutions. The outcome could determine whether future networks are truly autonomous or merely “AI-assisted.” For mobile operators struggling with profitability amid rising energy costs, the promise of self-optimizing networks represents both an operational necessity and potential competitive advantage.
Implementation Challenges and Realistic Timeline
While the vision is compelling, the path to deployment faces significant hurdles. Regulatory compliance represents a major challenge, as autonomous network reconfiguration must not interfere with emergency services or violate spectrum licensing agreements. Security is another critical concern—distributed AI systems create a larger attack surface for potential bad actors. The three-year timeline suggests we’re looking at proof-of-concept demonstrations rather than immediate commercial deployment. Realistically, we won’t see fully autonomous networks until the 6G era, which is still 5-7 years away. However, incremental benefits in energy efficiency could begin appearing within existing 5G-Advanced networks as early as 2026, particularly in high-density urban environments where energy costs are most pressing.
Broader Impact Beyond Telecom
The research emerging from Trinity College Dublin and its partners in Ireland could have implications far beyond mobile networks. The same distributed AI architectures could revolutionize smart grid management, industrial IoT systems, and even autonomous vehicle coordination. The fundamental challenge of making distributed systems intelligent, efficient, and reliable applies across multiple domains. If successful, Naira could establish Ireland as a hub for sustainable AI infrastructure research, attracting further investment and talent to the region. More importantly, it could provide a blueprint for how we build intelligent infrastructure that serves human needs without consuming unsustainable amounts of energy—a challenge that extends well beyond telecommunications into nearly every aspect of modern digital life.
