Jun 2026· IEEE Conference on Network Softwarization· pp. 428-434· 0 citations· 15 references
Computer Science
Abstract
The evolution towards 6G edge-cloud ecosystems demands autonomous, intent-based network management to handle unprecedented infrastructure complexity. While Large Language Models offer promising capabilities for translating high-level user intents into network configurations, current monolithic approaches suffer from cognitive overload, hallucinations, and a profound inability to safely execute low-level data plane mutations. To bridge this gap, we introduce Edgent, a novel framework that integrates hierarchical Agentic AI with Extended Berkeley Packet Filter technologies via the Model Context Protocol. Edgent utilizes a state-driven Supervisor, enhanced by Retrieval-Augmented Generation, to decompose abstract human intents into deterministic execution graphs and dynamically delegate tasks to domain-specific worker agents. We empirically validate the framework by autonomously deploying a distributed, in-kernel DDoS mitigation pipeline across scaled containerized topologies containing up to 85 nodes. Extensive evaluations demonstrate high orchestration reliability; notably, even heavily quantized Small Language Models (e.g., 4B parameters) achieve near-perfect zero-shot execution and 100% overall task completion through autonomous error recovery. Finally, latency and resource profiling confirm that the multi-agent framework can be efficiently driven by fully localized models compatible with orchestration tasks directly within resource-constrained edge environments, therefore this work positions Edgent as a pragmatic step toward the realization of zero-touch nextgeneration networks.
Autonomous Agentic AI extends conventional Internet of Things (IoT) intelligence from isolated inference toward goal-directed systems that perceive, maintain state, plan, invoke services or actuators, coordinate with other agents, and adapt from feedback. Deploying such capabilities across device, edge, fog, and cloud tiers creates a coupled systems problem: cognitive architecture, organizational topology, workload placement, state management, and trust controls must be designed jointly. This systematic review maps and synthesizes peer-reviewed research on agentic and multi-agent architectures, edge intelligence, computation offloading, and continuum orchestration. A reproducible search of OpenAlex used 20 concept queries and 19 foundational-title lookups, followed by metadata screening and DOI-level Crossref verification. From 765 unique records, 251 entered detailed screening, 155 passed topical and quality criteria, and 153 peer-reviewed studies published between 2008 and 2026 formed the final corpus. The evidence was coded by agent paradigm, continuum tier, orchestration mechanism, evaluation maturity, and primary contribution. Agentic or multi-agent architecture was the largest evidence stream (67 studies; 43.8%), while optimization (82; 53.6%), reinforcement learning (44; 28.8%), and multi-agent reinforcement learning (40; 26.1%) were the most frequent orchestration mechanisms. However, 62 studies (40.5%) relied primarily on simulation, whereas only one (0.7%) reported an operational or pilot deployment. The synthesis yields a four-tier reference architecture with a cross-cutting control and assurance plane, a five-family orchestration taxonomy, and a deployment decision framework. 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. Standardized benchmarks, stateful-agent migration protocols, interoperable capability descriptions, and longitudinal operational evidence remain urgent research priorities.
The Organic 6G vision of a network of networks spanning an edge-cloud continuum complemented by non-terrestrial resources requires, to realize its promise, service provisioning that is simple to operate, scalable across independently administered domains, and agile under domain churn (i.e., domains dynamically joining and leaving). Despite advances in cross-domain orchestration, many proposals rely on heavy integration fabrics, multi-layer coordinators, and deep telemetry pipelines that hinder deployability and amplify coordination overhead. We propose a lightweight, decentralized conversational orchestration framework based on Large Language Model (LLM)-driven domain agents. Each domain remains autonomous: an agent observes local state via tools, reasons in a closed loop, and exchanges summaries with neighboring agents over an Agent-to-Agent (A2A) overlay aligned with data-plane coupling. Fast feasible placement is enabled by periodic, routing-like dissemination of reachability advertisements (latency, bottleneck bandwidth, and compute capacity), while safe re-optimization, scaling, and migration are handled through event-driven requests and negotiation. To meet real-time constraints, we deploy a compact reasoning model trained with verifier-based self-verification and periodically refined online via shadow updates. Simulations show manageable, near-linear control-plane overhead as domains scale and during domain joins, and robust decision quality, including recovery after objective changes. We close by outlining future research directions for principled, secure, and uncertainty-aware agentic orchestration in Organic 6G.
Large Language Models (LLMs) are increasingly deployed as autonomous agents, transitioning from static conversational interfaces to dynamic systems capable of complex reasoning, tool execution, and decision-making. However, the operational reliability of these agentic AI systems is fundamentally challenged by the absence of reliable ground truth in open-ended environments and the risk of increasing operational drift over time. To address this challenge, we propose and experimentally evaluate an agentic AI framework, designed to enforce autonomous integrity within LLM-driven systems. We design a self-calibration mechanism that mitigates drift and dynamically approximates ground truth by incorporating an ARIMA forecaster, without requiring continuous human oversight. To demonstrate the effectiveness and reliability of our methodology, we apply it to the complex domain of profiling the resource usage of zero-knowledge workloads in edge computing networks. Experimental results show that the proposed self-calibrating agentic framework successfully profiles the zero-knowledge workloads, achieving a higher accuracy than baseline LLM agents by 91.7% for resource usage prediction and improving the prediction speed by 71.7% compared to pure profiling, establishing a robust foundation for deploying autonomous AI in decentralized infrastructures. Furthermore, the ground truth generation using the proposed ARIMA leaping algorithm is 52% faster than a standard ARIMA forecasting algorithm, while achieving the same accuracy.
Fin Gentzen, Marla Grunewald, Iulisloi Zacarias et al.· 0 citations
The fast development of Edge AI has moved the intelligence towards distributed Edge-Cloud computing infrastructure moving away from centralized cloud in terms of low-latency, privacy-preserving, and context-aware applications. However, the deployment and management of the AI-based workloads in heterogeneous and resourceconstrained environments faces mounting complexity with the need for continuous and autonomous orchestration mechanisms. This paper provides a survey on Edge AI orchestration along the Edge-Cloud continuum based on a closed-loop lifecycle encompassing monitoring, decision making, placement, scaling, migration, and adaptation. We examine the ways in which modern systems sense dynamic environments, reason over multi-objective constraints, and perform runtime control actions to adapt and optimize performance, efficiency, and quality of service. Additionally, we systematically review state-of-the-art approaches starting from classic optimization, to emerging federated intelligence, autonomous agents, and foundation model-driven paradigms. Lastly, we identify key open challenges, including cross-layer monitors, explainable decision making, and continual adaptation under concept drift, outlining future research directions oriented towards fully autonomous and selfoptimizing Edge AI systems.
Dragan Stojanović, N. Stojanović, Aleksandra Stojnev· International Scientific Con...· 0 citations
Autonomous agents challenge conventional LLM serving by coupling repeated inference with persistent context and sandboxed tool execution. We present Aries, a full-stack experimentation framework that separates task semantics from execution configurations, reconstructs cross-component agent trajectories with correlated system telemetry, and exposes stateful tool execution through a consistent interface across heterogeneous sandbox substrates. We use Aries to conduct reproducible experiments on open agent harnesses and benchmarks. We complement these experiments with production traces from a commercial platform, grounding low-level systems research in observed production behavior. Our results show that (1) token-centric metrics miss non-inference bottlenecks, (2) retaining additional context yields diminishing accuracy benefits while reducing serving capacity, and (3) tool sandboxes alternate between long idle periods and short resource bursts, while current snapshot-based state management makes aggressive suspension costly. A complementary security analysis further highlights the need to reduce the sandbox attack surface. We then discuss the vision for agent-native serving systems designed around trajectory-level metrics, adaptive context management, elastic sandbox resource management, and sandboxes with minimized attack surface.
Leonid Kondrashov, Hongrui Liu, JooYoung Park et al.· 0 citations
Orchestrating services across heterogeneous 6G edge-cloud infrastructures requires autonomous coordination systems managing distributed computational resources while satisfying Quality-of-Service (QoS) requirements. Recent advances in Large Language Models (LLMs) enable development of autonomous agents capable of complex reasoning and decision-making for such orchestration tasks. However, applying generic agentic AI frameworks from the machine learning literature to orchestration domains introduces reliability limitations, as trial-and-error decision patterns are unsuitable for environments where errors disrupt services. This work presents AgentEdge, a novel distributed intelligence framework that implements specialized autonomous agents in four orchestration roles: intent processing, infrastructure monitoring, strategic planning, and action execution. AgentEdge introduces the PARES (Perceive, Act, Reason, Evaluate, Sustain) framework establishing minimum capabilities required for autonomous agent qualification. Central to AgentEdge is the ActSimCrit (Action-Simulation-Critic) planning methodology, which validates orchestration plans through digital twin simulation before execution, eliminating direct infrastructure experimentation risks. Agents coordinate multi-step operations and adapt strategies based on constraint feedback. Structured outputs constrain agent decision spaces to feasible orchestration actions while preserving optimization flexibility. Experimental evaluation in six orchestration scenarios validates AgentEdge through comparison with baseline agentic frameworks and ablation studies. AgentEdge achieves $2.76\times $ higher success rate compared to generic agentic frameworks (ReAct, LATS) and $10\times $ reduction in API call variability. The core ActSimCrit digital twin component alone contributes $1.47\times $ success improvement when compared to direct planning without simulation. AgentEdge achieves significant power savings across infrastructure scales from 8 to 35 nodes.
B. Gort, Godfrey M. Kibalya, A. Antonopoulos· IEEE Transactions on Machine...· 0 citations