NET3 – AI-native 6G networks and their digital twins
Wednesday, 5 June 2024, 11:00-13:00, room Gorilla 4
Session Chair: Marco Gramaglia (Universidad Carlos III de Madrid, ES)
Federated Learning for Anomaly Detection in Open RAN: Security Architecture Within a Digital Twin
Yasintha Rumesh, Dinaj Attanayaka, Pawani Porambage and Jarno E. Pinola (VTT Technical Research Centre of Finland, Finland); Joshua B Groen and Kaushik Chowdhury (Northeastern University, USA)
The Open Radio Access Network (Open RAN) specifies the evolution of RAN with a disaggregated, open and intelligent architecture to meet the requirements of next-generation networks. While this provides flexibility and optimization for RAN, it raises new security concerns, potentially increasing vulnerability to cyber threats through disaggregated elements. We introduce a security architecture that functions as a platform to evaluate configurations and train security algorithms within a Network Digital Twin (NDT), which is compliant with the O-RAN architecture defined by the O-RAN Alliance. The elements of the security architecture reside within the NDT and facilitate the training of machine learning (ML) models, which play a pivotal role in the O-RAN security operations. To exemplify this framework, we demonstrate a hierarchical Federated Learning (FL) based anomaly detection algorithm that can be applied for three traffic slices in O-RAN. We use Colosseum, an O-RAN-compliant emulation system, to generate time-series data for training. Our trained model is able to detect anomalous traffic and identify traffic slice types with over 99% accuracy.
Integrating Network Digital Twinning into Future AI-Based 6G Systems: The 6G-TWIN Vision
Sébastien Faye (Luxembourg Institute of Science and Technology (LIST), Luxembourg); Miguel Camelo Botero (IMEC, Belgium); Jean-Sébastien Sottet (Luxembourg Institute of Science and Technology, Luxembourg); Christoph Sommer and Mario Franke (TU Dresden, Germany); Julien Baudouin (Proximus Luxembourg, Luxembourg); German Castellanos (Accelleran NV, Belgium); Régis Decorme (R2M Solution, France); Maria Pia Fanti (Polytechnic of Bari, Italy); Ramin Fuladi (Ericsson & Boğaziçi, Turkey); Gunes Kesik (Ericsson, Turkey); Beatriz Mendes (Ubiwhere, Portugal); Chris Murphy (VIAVI Solutions, United Kingdom (Great Britain)); Stephen Parker and Simon Pryor (Accelleran NV, Belgium); Sidi-Mohammed Senouci (University of Bourgogne – ISAT Nevers, France); Ion Turcanu (Luxembourg Institute of Science and Technology, Luxembourg)
As we move closer to the 6G era, the complexity and dynamism of communication networks is set to increase significantly. This requires more sophisticated, automated management solutions, while retaining human oversight. To meet these new challenges, it is essential to build new architectures that provide a secure environment for closed-loop network automation and to enable optimal resource planning, management and control. In this context, this paper presents the vision of the European 6G-TWIN consortium, which focuses on the creation of a cyber-physical continuum that seamlessly merges the physical world with its digital representation, through the concept of Network Digital Twin (NDT) and its integration into an Artificial Intelligence (AI)-based 6G architecture. This initiative aims to bridge the gap between the increasing complexity of networks and their operational performance, potentially improving real-time user experiences across multiple domains. This paper specifically highlights the associated key challenges and the approach taken by 6G-TWIN to address them.
Security Framework in Digital Twin for O-RAN
Dinaj Attanayaka, Yasintha Rumesh, Jarno E. Pinola and Pawani Porambage (VTT Technical Research Centre of Finland, Finland)
With the vision of 6G networks, the Radio Access Networks (RAN) are expected to entail increased programmability, efficiency, and flexibility. The current trends in Open RAN architecture that promote openness, disaggregation, and inherent intelligence are well aligned with these aspects. The Open RAN architecture’s highly dynamic nature and the data-driven approaches necessitate more cognitive and proactive security measures with robust defensive mechanisms. Digital Twin (DT) is proven to be an ideal platform to integrate into such dynamic systems for testing and optimizing configurations and security algorithms. In this paper, we propose a DT-based security framework for Open RAN (O-RAN) with an elaboration of a use case of security service level agreement (SSLA) assurance management. Moreover, we describe how to integrate energy-aware solutions, such as green security policies, within the proposed framework.
A Human-In-The-Loop Based ML Framework to Estimate User’s QoE on Cloud Gaming Using Active Learning
Marcos Carvalho, Daniel H. C M. Soares and Daniel Fernandes Macedo (Universidade Federal de Minas Gerais, Brazil)
Data-driven techniques such as supervised machine learning demand a large number of labeled instances to generate precise models. On the other hand, frequently nagging users to provide inputs is not an option, since it reduces the user’s engagement with a service. This paper explores active learning to create a user (human) in the loop learning process: learning occurs in steps, where in each step the technique suggests new instances to be labeled by humans. We evaluate our proposal in an emulated network, where users label QoE for a cloud gaming application. Results show that the number of instances required to achieve a certain level of precision is reduced by 38.4%, while the model error decreases 37.1%.
AI-Native Architecture for 6G Networks and Services with Model Dependencies
Nassima Toumi and Toni Dimitrovski (TNO, The Netherlands)
Next generation networks and services increasingly rely on AI/ML methods to integrate intelligence in the decision making process with the introduction of MLOps pipelines for automating orchestration and lifecycle management. Further, Digital Twinning is viewed as a key enabling technology for 6G networks due to its foreseen benefits in network management optimization by providing a digital replica of the network. However, further work is required for an AI-native 6G architecture that fully integrates AI/ML into network and service management. Indeed, the use of Digital Twins and Transfer Learning leads to dependencies between models, which ought to be taken into account during their lifecycle management. Hence, the work described in this paper provides a framework for enabling an AI-native network architecture which supports the handling of AI/ML model interdependencies and relationships in network and service management and orchestration.
A Proof of Concept Implementation of an AI-Assisted User-Centric 6G Network
Nikolaos Gkatzios (INFOLYSiS P. C., Greece); Harilaos Koumaras (NCSR Demokritos, Greece); Dimitrios Fragkos (National Centre for Scientific Research Demokritos (NCSRD) & University of Peloponnese, Greece); Vaios Koumaras (INFOLYSiS P. C., Greece)
The design of the 6G system will be based on a user-centric paradigm, enabling users to be involved in the creation and the management of network services and also providing them with a highly customized network experience. This paper proposes a realization of this paradigm by proposing a core network redesign from “Network Function-focus” to “User-focus” in conjunction with an AI-assisted approach that self-organises the network according to the user-requirements. A proof of concept implementation is presented based on both simulated and physical deployments, that demonstrate an optimal UPF placement taking into consideration user’s preferences, proving the validity of the proposed approach.
The Role of AI Enablers in Overcoming Impairments in 6G Networks
Merve Saimler (Ericsson Research, Turkey); Selim Ickin (Ericsson Research, Sweden); Giacomo Bernini (Nextworks, Italy); Nassima Toumi (TNO, The Netherlands); Maria Diamanti (Institute of Communication and Computer Systems (ICCS) – National Technical University of Athens, Greece); Symeon Papavassiliou (National Technical University of Athens, Greece); Özgür Umut Akgül (Nokia, Finland); Bahare Masood Khorsandi (Nokia, Germany); Milan Zivkovic (Apple Technology Engineering B.V. and Co. KG, Germany)
The integration of Artificial Intelligence (AI) into the 6G architecture, referred to as AI-native 6G architecture, signifies a transformative era for communication technology. Nevertheless, practical implementation encounters challenges including architectural complexities, data quality concerns, and operational difficulties in managing machine learning models, allocating resources, and implementing intent-based management. In this paper, we present a comprehensive approach to address these challenges in emerging 6G networks through AI. Our approach involves two steps: first, we identify impairments hindering progress, analyzing the importance of addressing operational challenges in Machine Learning Operations (MLOps), 6G evolution, and democratizing AI, while addressing interoperability issues and complexities in the translation of business intents into network configurations. Upon the analysis, we highlight AI enablers—architectural enhancements, MLOps, Data Operations (DataOps), AI as a Service (AIaaS), and intent-based management— as essential solutions for practical AI implementation in 6G networks. We conclude by stating that architectural improvements prioritize privacy, security, and data accuracy, while MLOps and DataOps optimize the management of the AI life cycle. Privacy-aware data collection and training employ federated learning and split learning, and AIaaS streamlines AI access, and intent based management with integrated AI enhances decision-making through advanced algorithms.
