The study provides initial evidence of feasibility while identifying the challenges that must be addressed before production deployment and formalize the ADN agent model and workflow and define an operational framework covering communication, lifecycle management, governance, and security.
Abstract
Network intelligence has largely evolved around logically centralized control and orchestration. Although this model simplifies coordination, it creates a critical dependency on centralized services and limits localized adaptation. This paper presents Agentic-Defined Networking (ADN), an architecture that treats autonomous Artificial Intelligence agents as first-class entities embedded across the network infrastructure. ADN has two defining properties. First, agents perceive local state, maintain beliefs, reason over operator-defined objectives, coordinate with peers, and actuate programmable resources without requiring a persistent central controller in the critical decision path. Second, the mapping between agents and infrastructure is a deployment choice, supporting device-level, cluster-level, and hierarchical configurations. We formalize the ADN agent model and workflow and define an operational framework covering communication, lifecycle management, governance, and security. We evaluate feasibility and scaling through Mininet-AI experiments with topologies of up to 200 switches. The results characterize routing throughput, reasoning latency, fault-mitigation time, and coordination cost under distributed and hierarchical configurations. The study provides initial evidence of feasibility while identifying the challenges that must be addressed before production deployment.
Software-Defined Networking (SDN) has revolutionized network management by decoupling control logic from data forwarding. However, limited by the traditional controller paradigm, existing SDN controllers remain inherently static, relying on predefined rules. This rigidity makes them ill-equipped to handle unforeseen traffic patterns or emerging threats, often defaulting to generic actions that fail to address nuanced scenarios. Large-Language Models (LLMs), a group of models with billions of parameters that are trained on diverse datasets, are known to excel at performing complex tasks that require human-level reasoning and prior knowledge. With such powerful models assumed to encapsulate the collective knowledge of network operations within their parameters, one question that this work asks is "Are static networking rules provided by humans or by heuristics still relevant?" To answer this question, we propose a new logically centralized controller powered entirely by an LLM, called Agentic-Defined Networking (ADN). ADN introduces a novel architecture that integrates LLMs as the reasoning core of an SDN control plane implemented in a real network controller. To support the main thesis, we present preliminary results on ADN's performance on dealing with unseen malicious traffic and congestion-aware routing.
Shanaya Varkey, Sean Choi· Proceedings of the ACM SIGCO...· 0 citations
The main conclusion is that practical Agentic IoT depends less on placing an entire agent at one tier than on partitioning perception, memory, reasoning, and action under explicit latency, privacy, reliability, and safety constraints.
The rapid development of agentic AI and multi-agent systems is establishing AI agent communication as a fundamental requirement for the future Internet. While a diverse array of agent communication protocols has recently emerged, these solutions currently suffer from interoperability crises and infrastructure gaps. The newly proposed Service-Oriented Virtualization-Based Architecture (SOVA) offers an architectural framework to address these challenges for agent communication, which expects seamless support from the network infrastructure. The emerging AI-native 6G network is promising as a robust foundation for the SOVA framework, thereby greatly facilitating AI agent communication; however, its effectiveness in supporting the SOVA framework has yet to be fully assessed. To bridge the distinct research trajectories of AI-native 6G networks and AI agent communications, this paper investigates the capabilities of current and proposed 6G network architectures and protocol specifications for supporting the SOVA framework for AI agent communications. By critically examining 6G's key architectural paradigms and their potential to fulfill SOVA's requirements, this paper identifies gaps between 6G standards and the demands of AI agent communication. Based on this gap analysis, this paper outlines research and development directions to ensure that the future 6G network can natively empower AI agent communications in the era of agentic AI.
Distributed agentic artificial intelligence increasingly relies on networked groups of autonomous agents that exchange observations, intentions, plans and task states to achieve collective goals. This systematic review synthesizes peer-reviewed studies published between 2023 and 2026 on communication-efficient networking for distributed agentic AI, multi-agent reinforcement learning and networked autonomous systems. Following PRISMA 2020, studies were identified from major scholarly databases and analyzed thematically across communication protocols, message compression and selection, semantic communication, coordination mechanisms, latency reduction, energy efficiency and deployment challenges. The evidence shows that selective engagement, graph-based compression, delay-aware communication, implicit consensus and value-of-information scheduling can reduce redundant exchanges while maintaining coordination quality. Semantic and edge–cloud approaches further lower payload size and local computational demand, although their benefits depend on channel conditions, resource availability and task placement. Persistent limitations include scalability, protocol interoperability, security and privacy risks, inconsistent energy reporting, dependence on simulated environments and limited standardization of evaluation metrics. The review concludes that communication efficiency should be treated as a joint optimization problem involving bandwidth, latency, computation, energy and task performance. Future research should prioritize interoperable protocols, adaptive communication topologies, secure message exchange, realistic testbeds and standardized reporting frameworks for dependable, scalable and sustainable distributed agentic systems at scale.
Experimentation on distributed, heterogeneous computing environments—from edge devices to large-scale cloud platforms—demands orchestration technologies that are both flexible and extensible. Kiso is an open-source framework designed to provision resources and manage complex scientific workflows across the edge-to-cloud continuum. Its architecture unifies infrastructure provisioning, experiment configuration, and reproducible execution, enabling researchers to compose and monitor experiments that span geographically dispersed sites and variable network conditions. Although Kiso was conceived for workflow management—coordinating data-intensive tasks and ensuring reproducibility across dynamic infrastructures—its modular design makes it equally promising for providing reproducible environments for deploying and studying emerging agentic frameworks, where autonomous AI agents require consistent resource provisioning, cross-site communication, and result collection. We describe Kiso’s core capabilities for resource orchestration, experiment lifecycle management, and integration with containerized services, and we outline how these capabilities can support distributed multi-agent systems. In particular, we discuss how its declarative provisioning, extensible task abstractions, and built-in monitoring and output collection provide a natural foundation for experiments in which reasoning agents plan, negotiate, and adapt in real time. This study situates Kiso at the intersection of scientific workflow management and complex, agent-based computing, highlighting its potential to accelerate research on adaptive, self-organizing cyber-physical systems—an emerging frontier in complex systems science.
R. Mayani, K. Vahi, M. Rynge et al.· Frontiers in Complex Systems· 1 citation
A compact reasoning model trained with verifier-based self-verification and periodically refined online via shadow updates is deployed, showing manageable, near-linear control-plane overhead as domains scale and during domain joins, and robust decision quality, including recovery after objective changes.
Masoud Shokrnezhad, T. Taleb· IEEE Network· 0 citations