Deliverable D4.1: First design and implementation of control plane infrastructure
This deliverable presents the first design and implementation of the SEASON control plane, reporting the initial outcomes of all WP4 activities. It covers monitoring and telemetry, single- and multi-domain control, and AI/ML-based self-management for end-to-end transport networks.
The report first outlines WP4 objectives, KPIs, and the overall SEASON control-plane architecture, describing the key network components and control technologies across the full transport network, from RAN to core.
A major contribution is the design and preliminary implementation of a comprehensive monitoring and streaming telemetry framework. Network elements expose configuration and operational data to reflect real-time network state. Intelligent data aggregation algorithms are introduced to handle the 5 V’s of telemetry data (volume, velocity, variety, veracity, value), enabling scalable, efficient analytics and actionable insights for network automation.
The deliverable then details SDN-based control of access infrastructures, focusing on SDM-PON architectures. This includes centralized control of optical and spatial devices, such as OLTs and spatial aggregation/disaggregation elements, enabling flexible management of multi-fiber and multi-core optical scenarios.
For the transport network, a model-driven control approach is presented, relying on YANG-based management. OpenConfig models are used for transceivers and packet/optical devices, while OpenROADM models are employed for ROADMs, with additional evaluation of OpenROADM support for pluggables in packet/optical systems.
The report also introduces digital twin concepts for optical transport networks, enabling virtual representations of network elements that replicate real device behavior for configuration, monitoring, and service management. The proposed digital twin is evaluated using simulators, vendor emulators, and OpenROADM-based platforms. An Optical Time Domain Digital Twin is further presented for applications such as QoT estimation and failure management, supported by ML-based models.
Finally, the deliverable explores AI/ML-driven service management and orchestration, integrating techniques such as SVMs and DNNs to enable near real-time decision-making. These approaches support automated service orchestration, application placement, energy control, and multi-layer resource coordination across RAN, optical, transport, and core networks, targeting efficient service provisioning in B5G environments.
EASON Architecture overarching control for RAN/PON/Backhaul network segments