Organizations: KTH Royal Institute of Technology, Stockholm, Sweden · SSH Communications Security, Helsinki, Finland · Tech Validation–AI Department, Chief Technology Officer (CTO) Organization, Lenovo, Beijing, China · Uppsala University and RISE, Sweden
Agentic Artificial Intelligence (AI) and Network Digital Twins (NDTs) are increasingly combined to form Cognitive Digital Twins (CDTs) that translate operational intents into network changes and support closed-loop management. However, a network change may achieve its intended objective while disrupting the monitoring or recovery mechanisms needed for subsequent decisions. For example, a service may be rerouted while its latency-monitoring traffic continues along the original path, producing measurements that no longer reflect the service's performance. We propose Closed-Loop Dependency Assurance (CLDA), an assurance mechanism that augments agentic AI-enabled NDTs with dependency-aware validation. In our framework, an agentic AI planner generates and refines network plans, while the NDT evaluates the projected network outcomes of these plans. CLDA extends this analysis using a Loop Dependency Graph (LDG) that connects service requirements to monitoring mechanisms, forwarding paths, failure domains, and recovery prerequisites. CLDA projects candidate actions onto these dependencies, checks intermediate states of multistep plans, and supports deterministic make-before-break repairs using temporary monitoring mechanisms. Unlike the Large Language Model (LLM)-based planner, CLDA uses explicit, deterministic rules for assurance decisions and repair. We evaluate CLDA through controlled reconfigurations, a 240-trial LLM-based planning campaign, and live network execution. A separate synthetic study examines deterministic assurance scalability across network models representing up to 1,024 routers.
Figures & tables
Fig. 1: CLDA -enhanced agentic NDT architecture and assurance workflow. The agentic AI planner generates network plans from operator intents, the NDT projects their network effects, and CLDA evaluates assurance dependencies before accepting, repairing, or rejecting proposed changes.
Fig. 2: Planning and execution with CLDA . Candidate actions are projected before commit; during execution, the live state is refreshed after network-changing actions so the controller can continue, repair, stop, or recover.
Fig. 3: Outcomes of the evaluated assurance approaches on 15 targeted cases, comprising five unsafe and ten benign changes. Bars indicate the number of unsafe changes rejected and benign changes accepted, with counts relative to each category’s total.
Fig. 4: Outcome distribution across the 240 -trial agentic planning campaign. Each bar contains 30 trials from one intent family; planner-invalid and goal-mismatched outputs are separated from NDT and CLDA decisions.
Fig. 5: Operational cost and scalability of CLDA . (a) Representative prototype time scales, separating deterministic assurance from live network interaction and agentic planning. (b) Synthetic scaling from 8 to 1,024 routers ( 30 to 3,840 LDG nodes). Lines connect medians over 200 measured repetitions after 40 warm-ups; shaded regions extend to p95.
Ahmed Hussain is a Doctoral Researcher at KTH Royal Institute of Technology, Stockholm, Sweden. His research interests include IoT security and privacy, wireless security, Edge AI, Agentic AI, and AI-enabled cybersecurity. Website: https://ahmed-hussain.net .
Table 6
Konstantinos Kalogiannis is a PhD Candidate at KTH Royal Institute of Technology, Stockholm, Sweden. His research interests include security and privacy in cyber-physical systems, with a focus on vehicular ad hoc networks (VANETs) and platooning, and the security analysis of large-scale software systems.
Table 7
Salahuddin Salahuddin is currently an AI Researcher at SSH Communications Security. He received his M.Sc. through a double-degree program from Aalto University, Finland, and KTH Royal Institute of Technology, Sweden. His research interests include artificial intelligence, cybersecurity, machine learning, and the application of AI techniques to security-related challenges.
Table 8
Hexu Li is currently a Researcher working with the Tech Validation – AI department at the Chief Technology Officer (CTO) Organization, Lenovo, Beijing, China. He received his Master’s degree in Communication Systems from KTH Royal Institute of Technology, Stockholm, Sweden.
Table 9
Thiemo Voigt received the Ph.D. degree from Uppsala University, Sweden, in 2002. He is currently a Full Professor at the Department of Electrical Engineering, Uppsala University, and a Senior Researcher at RISE Computer Science. His current research interests include low-power networking, system software, and machine learning for embedded networked devices and the Internet of Things, as well as in-body communication and 6G Ambient IoT.
Autonomous network management requires systems that can evaluate Network Service Intents (NSIs) under varying conditions without manual implementation of analysis logic, as envisioned in TM Forum Level~4 (L4) autonomy. Behavioral Network Digital Twins (NDTs) enable such evaluation, but existing NDTs rely on pre-defined analytical logic, limiting adaptability for evolving closed-loop control. This paper introduces the NDT factory, a multi-agent software system that synthesizes executable behavioral NDTs on demand from semantic models using Large Language Model (LLM). We validate the system using a Call Admission Control (CAC) case study, where deterministic what-if analysis serves as the admission decision process. The NDT factory generates a complete CAC NDT through parallel synthesis and orchestration, achieving 100% compilation and test pass rates across multiple runs. Simulation over 300 NSIs shows 99.3% decision agreement with a reference implementation, 90% admission rate, and correct attribution of all rejections, demonstrating reliable synthesis with deterministic, verifiable execution.
Sudipta Acharya, Petar Djukic, Burak Kantarci
University of Ottawa, Ottawa, ON, Canada · Bell Labs Research, 600 March Road, Kanata, ON K2K 2E6, Canada
Network change validation remains a critical yet predominantly manual, time-consuming, and error-prone process in modern network operations. While formal network verification has made substantial progress in proving correctness properties, it is typically applied in offline, pre-deployment settings and faces challenges in accommodating continuous changes and validating live production behavior. Current operational approaches typically involve scattered testing tools, resulting in partial coverage and errors that surface only after deployment. In this paper, we present Aether, a novel approach that integrates Generative Agentic AI with a multi-functional Network Digital Twin to automate and streamline network change validation workflows. It features an agentic architecture with five specialized Network Operations AI agents that collaboratively handle the change validation lifecycle from intent analysis to network verification and testing. Aether agents use a unified Network Digital Twin integrating modeling, simulation, and emulation to maintain a consistent, up-to-date network view for verification and testing. By orchestrating agent collaboration atop this digital twin, Aether enables automated, rapid network change validation while reducing manual effort, minimizing errors, and improving operational agility and cost-effectiveness. We evaluate Aether over synthetic network change scenarios covering main classes of network changes and on past incidents from a major ISP operational network, demonstrating promising results in error detection (100%), diagnostic coverage (92-96%), and speed (6-7 minutes) over traditional methods.
Jordan Auge, Sam Betts, Giovanna Carofiglio +3
Cisco Systems · Cisco Systems, Paris, France and London, UK. · Swisscom +1
Network incident response remains slow and labor-intensive as the defender must infer multi-stage attacks from partial observations and translate recovery decisions into reliable system commands. Decision-theoretic planners provide principled optimization but typically rely on abstract states and predefined actions, while large language model (LLM) agents can reason over operational context but may hallucinate attacks and responses. Toward automating response planning, we present a hierarchical agentic response framework that integrates LLM-based attack inference, rollout planning, and digital-twin validation. A fine-tuned LLM infers the attack progression and affected hosts from security alerts and system measurements. An emulated network digital twin replays the inferred attack and returns discrepancies between predicted and observed effects to calibrate the inference. A separately fine-tuned planning agent uses the rollout planning method to prioritize affected components at the tactical layer. At the operational layer, the planning agent proposes high-level recovery actions, and an execution agent translates selected actions into recovery and verification commands that are validated in the digital twin. We evaluate the framework on a 33-component enterprise-network testbed under three multi-stage attack scenarios. The results show that our framework outperforms frontier-LLM baselines in recovery success rate by 18--31%.
Yiran Gao, Juntao Chen, Tao Li
Department of Systems Engineering, City University of Hong Kong, Hong Kong SAR, 99977 · Department of Computer and Information Sciences, Fordham University, New York, NY, 10023, USA