NET3 – Network Softwarisation
Thursday, 4 June 2026, 11:00-12:30, room Dirección de Certámenes (1st floor)
Session Chair: Aggelos Bletsas (Rutgers University & WINLAB, US)
Sustainable Reward Mechanisms for Multi-Objective Policy Design Using Reinforcement Learning in 6G Networks
Aamir Latif (Tallinn University of Technology, Estonia); Yannick Le Moullec (Tallinn University of Technology (TalTech), Estonia); Muhammad Mahtab Alam (Tallinn University of Technology, Estonia)
Sustainable sixth-generation (6G) wireless networks are expected to deliver massive connectivity, extreme energy efficiency, and ultra-reliability simultaneously, a triad of inherently conflicting objectives characterized by complex, non-linear trade-offs. Multi-Objective Reinforcement Learning (MORL) provides a principled framework to address such trade-offs; however, existing approaches typically rely on static scalarization or require training multiple policies for different preferences, limiting adaptability and practical deployment. Recent envelope-based policy optimization methods improve training stability under conflicting gradients. Yet, they remain restricted to fixed or externally specified preferences and do not generalize across the continuous trade-off space at inference time. This paper proposes a Wireless Aware Multi-Objective Conditional Envelope PPO (WAMO-CEPPO), a MORL framework that learns a single preference-conditioned policy that adapts online to varying network conditions without retraining. The framework integrates three key components: (i) a context-to-preference mapper (CPM) that infers objective priorities from real-time network states, (ii) a conditional envelope mechanism that mitigates gradient conflicts during policy updates, and (iii) a sustainable reward formulation that jointly accounts for energy efficiency, connectivity, reliability, interference, and hardware sustainability. Hardware sustainability is explicitly modeled to reflect computational efficiency and convergence behavior, aligning learning performance with green networking objectives. Simulations show that WAMO-CEPPO achieves up to 39% higher Pareto coverage, Hypervolume of 0.752 (+38.7% over MO-PPO), and IGD reduction of 58.7%, while improving reliability by up to 23%, EE by 19%, and fairness by 17% compared to state-of-the-art MORL baselines.
In-Network Cyber Protection for Modern EPES: The COCOON Programmable Node Architecture
Filip Holik (University of Glasgow, United Kingdom (Great Britain) & Norwegian University of Science and Technology, Norway); Simon Jouet and Bikram Singh Deol (University of Glasgow, United Kingdom (Great Britain)); Tahira Mahboob (University of Glasgow, United Kingdom (Great Britain) & Information Technology University, Pakistan); Awais Aziz Shah and Dimitrios P Pezaros (University of Glasgow, United Kingdom (Great Britain))
As part of the COoperative Cyber prOtectiON for modern power grids (COCOON) Horizon Europe Innovation Action, researchers at the University of Glasgow have built COCOON Programmable Node (CPN) architecture to support stateful end-host network stack programmability and function composition for high-speed cybersecurity monitoring and control in resource-constrained environments such as IoT, Edge, and industrial networks while being platform independent. This CPN architecture comprises an eBPF-based programmable data plane, CPN controller with pre-compiled eBPF micro network functions, and the application layer supporting high-level third-party applications implementing advanced cybersecurity functionality. The architecture is deployed in four real-world electrical power and energy systems pilot setups and validated against power grid network requirements using representative datasets. The validation results show that the CPN architecture is highly suitable for these networks and can be generalised and deployed within all critical infrastructure areas.
Telemetry-as-a-Service: Decoupling Collection from Injection for Scalable Network Monitoring
Carlos Natalino (Chalmers University of Technology, Sweden); Kaida Kaeval, Hendrik Johann Kerm and Torm Järvelill (Tallinn University of Technology, Estonia); Jasper Mueller (Adtran, Sweden); Paolo Monti (Chalmers University of Technology, Sweden)
Network telemetry has evolved from poll-based SNMP to push-based streaming protocols such as gNMI and YANG-Push, yet current architectures tightly couple device collection with pipeline injection inside monolithic collectors. Scaling telemetry processing under this model requires replicating device credentials across every collector instance, and outsourcing processing to third parties exposes management-plane access together with internal network topology. The term “telemetry-as-a-service” (TaaS) has been introduced in the literature as an API abstraction, but its architecture still combined collection and processing while retaining full device credentials. In this paper we extend the TaaS concept by separating stateful collectors, which face devices and hold credentials, from stateless injectors, which are credential-free, topology-agnostic, and horizontally scalable. The key enabler is a push-direction inversion: collectors push telemetry to injectors rather than injectors pulling from devices, creating a natural credential boundary. We present a gRPC-based protocol that preserves gNMI semantics while enforcing this boundary, and evaluate an implementation demonstrating linear throughput scaling with minimal latency overhead. The architecture enables independent scaling of collection and injection tiers, credential isolation suitable for outsourcing, and topology privacy preservation.
Intelligent Connectivity Platform: Intent-Driven Service Orchestration for Rural Networks
Muhammad Faheem Awan (Telenor Research and Innovation, Norway); Min Xie (Telenor Research & Telenor Group, Norway); Abdelhakim Cherifi, Jane Frances Pajo and Ali Esmaeily (Telenor Research and Innovation, Norway); Fernando Camacho (Huawei Technologies Co, Ireland); Lorenzo Cipriani (Huawei Technologies Ireland Research Center, Ireland); Kevin McDonnell (Huawei Technologies Co, Ireland); David Pradas (Viveris Technologies, France); Alejandro Ramírez-Arroyo (Aalborg University, Denmark); Maria Rita Palattella (Luxembourg Institute of Science and Technology (LIST), Luxembourg)
This paper presents the Intelligent Connectivity Platform (ICP) – an intent-based orchestration system for connectivity services in rural networks. The ICP uses a hybrid approach: it combines semantic analysis (via a Large Language Model) with KPI-based filtering to identify the most suitable connectivity solution for a given user request. The platform triggers network slice selection and resource reservation through standardised TM Forum Open APIs and interfaces seamlessly with existing service orchestrators. We validate the ICP in the framework of the Horizon Europe COMMECT project, with a real deployment at the Telenor Lab, where it translates user intents into standards-compliant 5G network services. This demonstration shows that ICP can automate service provisioning, improve resource allocation, and adapt to future network advancements.
Intent-Based Orchestration in Open RAN: An ns-3 Simulation Framework
Pouya Agheli (Orange, France); Grégoire Lefebvre (Orange Research, France)
This paper presents an extensible ns-3-based simulation framework for evaluating intent-based, semantics-aware control in Open RAN architectures. The framework integrates external Radio Access Network (RAN) Intelligent Controller (RIC) components and supports fine-grained control via internal distributed applications (dApps), enabling intent-based RAN orchestration across different timescales while maintaining standardized network behavior. As an illustrative use case, we implement an intent-based dApp for radio resource management (RRM) under realistic observability constraints. The scheduling problem is formulated using realistic key performance measurements (KPMs) available to dApps, together with a newly introduced Intent Satisfaction Score (ISS), which quantifies the delivery of intent-relevant information by combining distortion- and perception-oriented measures. Simulation results show that intent-based RRM can significantly improve ISS while reducing radio resource usage, at the cost of lower packet delivery ratio and throughput, and of moderate computational overhead.
