A unified service‐based capability exposure framework for closed‐loop network automation

Abstract

The ongoing quest for the tight integration of network operation and the network service provisioning initiated with the introduction of 5G often clashes with the capacity of current network architectures to provide means for such integration. Owing to the traditional design of mobile networks, which barely required a tight interaction, network elements offer capabilities for their continuous optimization just within their domain (e.g., access, or core), allowing for a "silo-style" automation that falls short when aiming at closed-loop automation that embraces all the actors involved in the network, from network functions up to the service-provider network functions. To this end, in this paper we make the case for the network-wide capability exposure framework for closed-loop automation by ( ) defining the different entities that shall expose capabilities, and ( ) discussing why the state of the art solutions are not enough to support this vision. Our proposed architecture, which relies on registration and discovery, and exposure functions, allows for enhanced use cases that are currently not possible with state of the art solution. We prove the feasibility of our solution by implementing it in a real-world testbed, employing Artificial Intelligence algorithms to close the loop for the management of the radio access network.

Adopting Artificial Intelligence Technology for Network Operations in Digital Transformation

“…They also pointed out that critical factors in this respect include data, leadership competency, and strategy and that attribute factors include data availability and quality, strategies and business needs, customer readiness and support, AI implementation and utilization capabilities, and cooperation and communication between organizations. In addition, Gramaglia et al (2022) analyzed the success factors of AI adoption, emphasizing the importance of support from the top management, technology capabilities, data, budgets, and employees' roles and abilities to implement AL algorithms and interpret the results. Pillai and Sivathanu (2020) stated that the top management's support is a deciding factor, while other factors, such as technological capability, external support and competition pressure, are slightly influential.…”

Management Closed Control Loop Automation for Improved RAN Slice Performance by Subslicing

“…While we propose a CCL implementation for subslicing, the full CCL is implemented in [12] to mitigate connection loss for moving UE. UEs are ships in real-world seaport testbed.…”

ATHENA: Machine Learning and Reasoning for Radio Resources Scheduling in vRAN Systems

“…Initiatives for including Artificial Intelligence (AI) and Machine Learning (ML) in the Radio Access Network (RAN) [1], Core [2], and Management [3] have been studied by the 3rd Generation Partnership Project (3GPP), while other initiatives led by the European Telecommunications Standards Institute (ETSI), such as the Experiential Networked Intelligence (ENI) [4] and Zero-touch Network and Service Management (ZSM) [5] groups, and industrial consortia such as O-RAN [6] are also promoting the introduction of data analytics into their architecture [7], [8], [9]. This integration will create control loops [10] among different network domains (e.g., RAN / Core Management) to seamlessly operate mobile networks, harvesting data from the different available sources (i.e., network functions, infrastructure), and enforcing back automated decisions into it. Integrating the application of data analytics tasks and intelligent algorithms into the network architecture could benefit the autonomous operation of the system, where components are self-organizing and self-orchestrating because it promotes the automation of decisions within the network.…”