Trust-based intent management for 6G: A level of trust assessment function

• Trust as a new intent for orchestrating the future 6G networks. • Continuous network assurance for cloud continuum services. • The Level of Trust framework monitors and double checks trust level agreements. • Trust-oriented ontology for Cloud Continuum and intent-based scenarios. Intent-Based Networking (IBN) has gained prominence from both industry and research communities for boosting network automation and reducing network complexity in multi-stakeholder scenarios. IBN helps service and network providers understand and translate high-level business goals by escaping from technical details. Nevertheless, the rapid evolution of service requirements in dynamic multi-stakeholder environments needs to be considered by suppliers in order to meet the continuous demands of their customers. In this regard, trustworthiness is recognized as a key feature to assess reliability, compliance, and user perception of 6G services. In that spirit, this article aims to pave the way for encompassing trustworthiness, or Level of Trust (LoT), as a new intent to properly deliver service provisioning in 6G solutions. In particular, the LoT framework introduces a set of functionalities to guarantee a proper definition of requirements (Trust Level Agreements, TLAs), their interpretation, translation, consistency, and assurance during an ongoing relationship. To this end, this article presents a trust-based intent management approach with an ad-hoc interpreter that is, in turn, powered by a trust-based ontology for Cloud Continuum scenarios. Besides, a Level of Trust Assessment Function (LoTAF) is designed to monitor and report updates and events regarding LoT continuously. Such a monitoring engine follows the IETF basics of service assurance for IBN architecture. Last but not least, experiments in a collaborative robot warehouse scenario showcase that LoTAF enables workload allocation to the most trustworthy compute nodes, improving confidence in service orchestration, while introducing negligible overheads (1.40 % RAM, 5.01 % CPU, and 0.1069s latency).