IBON will design and develop a novel smart end-to-end platform targeting network (from the RAN to the transport) and computing (from edge to cloud) self-optimization to provide committed Quality of Experience (QoE) to intelligent applications. The solution is based on:

1. Secure and Explainable AI-based IBN solution: In IBON, AI will be ubiquitous; intents are not only based on accurate AI models, but also they can exchange knowledge and be collectively considered in the decision making process to provide network services with tight coordination and assurance automation. Human operators can be involved in the decision-making process (operator-in-the-loop) and therefore, AI explainability must be addressed. Different heterogenous solutions to enforce data and AI model integrity and confidentiality will be federated into one fail-safe overarching centralized AI security enforcer delivering tokens, itself protected by trusted execution.

2. Zero-touch control and orchestration: The IBON architecture will be built on top of recent standards and open source projects; it will be hierarchical, with an e2e service orchestrator coordinating non-real time (non-RT) orchestrators for the RAN, transport network (packet and optical), and edge and core cloud computing and storage. The focus is on the transport network, where the orchestrator coordinates several near-RT Software Defined Network (SDN) controllers, which are placed close to the data plane network elements. A key innovation is the AI Function Orchestrator (AIFO), which is an independent orchestrator that will manage AI pipelines by placing functions in different domains and locations in the network and manage their connectivity to create a chain. Model training will be performed in sandbox domains with data from a Data Lake populated from heterogeneous data and context sources, including network, applications, and other systems, and augmented with data from simulation.

3. Adaptive network operations and service assurance: Network and services proactive adaptation and reconfiguration for service assurance will be based on AI prediction built on observed patterns and near optimal e2e resource allocation. Pervasive Quality of Transmission (QoT) and Quality of Service (QoS) telemetry data will feed AI algorithms for rapid detection of anomalies and performance degradations, which will make the data plane highly predictable and reliable. IBON will work on strategies for making autonomous decisions at the right level (node, controller, and/or orchestrator) and they might include operator-in-the-loop based on defined policies.

4. Intelligent applications will benefit from the targeted platform, as it will enable making dynamic resource adaptation, including the placement of virtual functions and connectivity services, based on Key Quality Indicators (KQI) (elaborated through end-to-end QoT/QoS telemetry, context and other meta-data), for perceived zero latency and application-level resilience, and reach a superior QoE.