Advanced networking, computing and software technologies are being developed to efficiently serve the strict requirements of the heterogeneous 5G and beyond application scenarios. The respective infrastructures are undergoing a major transformation towards a disaggregated model, with the adoption of Software Defined Networking, Network Function Virtualization and cloud-native principles that enable the virtualization and easy deployment of various services over rapidly allocated network slices.

Nevertheless, the plethora of the current and future verticals and uses cases requires an abstraction layer to be in place, which will automate their efficient deployment by fully exploiting the available technologies. This layer will be part of a network agnostic automation process that translates applications’ high-level requirements to particular configuration parameters that are then applied by SDN to individual equipment. In this way, the overall infrastructure configuration will be performed in a simple, holistic and integrated manner, overcoming the barriers set by the increased complexity of the heterogeneous set of wireless and optical networking and computing resources, deployed in the fronthaul, midhaul and backhaul network segments.

Towards this direction, in MAESTRO we propose a novel intent-based architecture for the orchestration of converged fiber-wireless-computation-storage infrastructures, adopting a closed-loop structure for service lifecycle management. The proposed optimization mechanisms will be based on Machine Learning, Artificial Intelligence, Deep-Reinforcement Learning and multi-objective optimization methods to continuously improve application service delivery, a challenging aspect considering the heterogeneity of the network.

MAESTRO enables the automation of an infrastructure’s configuration, translating intents to resource constraints, deciding on the resource allocation and function/service workload placement, monitoring in real-time through telemetry the infrastructure and applying configuration changes programmable.