| Innovative 5G orchestration architectures so far, have been mainly designed and optimized for Quality of Service (QoS), but are not aware of Quality of Experience (QoE). This makes intent recognition and End-to-End interpretability an inherited problem for orchestration systems, leading to possible creation of ineffective control policies. In this paper, an AI-driven intent-based networking for autonomous robots is proposed and demonstrated through the 5G-ERA project. In particular, to map an intent from individual vertical action to a global OSM control policy, a workflow of four tools is proposed: i) Action Sequence Generation, ii) Network Intent Estimation, iii) Resource Usage Forecasting, and iv) OSM Control Policy Generation. All of these tools are described in the paper with specific function descriptions, inputs, outputs and the semantic models/Machine Learning tools that have been used. Finally, the paper presents the developed intent-based dashboard for the visualization of the tools’ outputs, whilst taking QoE into consideration. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.