Leveraging AI Agents for Autonomous Networks: A Reference Architecture and Empirical Studies

2025-09-10

AI Summary

The TM Forum’s Level-4 Autonomous Network (AN) vision—encompassing self-configuration, self-healing, self-optimization, and zero-wait, zero-touch, zero-failure service—remains hindered by latency bottlenecks in traditional reactive automation. Method: This paper proposes and validates an end-to-end reference architecture based on cognitive AI agents, integrating proactive reasoning with reactive control. It adopts the Joseph Sifakis agent framework, hybrid knowledge representation, and RAN link adaptation techniques to enable real-time decision-making. Contribution/Results: Deployed in a 5G NR sub-6 GHz testbed, the system achieves sub-10 ms dynamic modulation and coding scheme (MCS) decisions, yielding a 6% increase in downlink throughput and a 67% reduction in block error rate—significantly enhancing ultra-reliable low-latency communication (URLLC). Crucially, this work pioneers the first practical deployment of coordinated cognitive agents for closed-loop, real-time control in the wireless access network, establishing a scalable and empirically verifiable pathway toward L4 autonomy.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Multiagent Systems: Agent Communication
• Machine Learning: Hardware-aware ML

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Agentic search

Abstract

The evolution toward Level 4 (L4) Autonomous Networks (AN) represents a strategic inflection point in telecommunications, where networks must transcend reactive automation to achieve genuine cognitive capabilities--fulfilling TM Forum's vision of self-configuring, self-healing, and self-optimizing systems that deliver zero-wait, zero-touch, and zero-fault services. This work bridges the gap between architectural theory and operational reality by implementing Joseph Sifakis's AN Agent reference architecture in a functional cognitive system, deploying coordinated proactive-reactive runtimes driven by hybrid knowledge representation. Through an empirical case study of a Radio Access Network (RAN) Link Adaptation (LA) Agent, we validate this framework's transformative potential: demonstrating sub-10 ms real-time control in 5G NR sub-6 GHz while achieving 6% higher downlink throughput than Outer Loop Link Adaptation (OLLA) algorithms and 67% Block Error Rate (BLER) reduction for ultra-reliable services through dynamic Modulation and Coding Scheme (MCS) optimization. These improvements confirm the architecture's viability in overcoming traditional autonomy barriers and advancing critical L4-enabling capabilities toward next-generation objectives.

Problem

• Achieving cognitive capabilities in autonomous networks
• Bridging architectural theory with operational reality
• Overcoming traditional barriers to network autonomy

Innovation

• Implemented cognitive system with hybrid knowledge representation
• Deployed proactive-reactive runtimes for autonomous network control
• Achieved dynamic MCS optimization for 5G throughput improvement