Page 188 - AI for Good-Innovate for Impact

AI for Good-Innovate for Impact

the radio network to simulate location and radio conditions. RAN metrics (such as channel
conditions, cells, bands, resources), 5G Core metrics (including network functions location,
resources, and configurations), MCX metrics (like users, groups, calls, bytes transmitted/
received, codecs), and equipment metrics (such as CPU, disk, memory, and network usage)
would be included. Interaction through an interface for real-time/near-real-time metrics retrieval
would be facilitated by the environment.

The simulated data would then be analyzed as a time series from a cumulative database and
as a Reinforcement Learning agent-environment for optimal policy learning. Adaptation of
the 5G network for KPI violations, error detection, or capacity bottlenecks would be pursued.
Predictions on KPI violations, error detection, or capacity bottlenecks, along with optimal policy
triggers, would be demonstrated and published based on the simulated data patterns.

43�3� Use case requirements

• REQ-01: It is critical that the solution/system enables autonomous network operations
through ML/AI to simplify deployment and configuration tasks.
• REQ-02: It is critical that the solution/system incorporates a Digital Twin for efficient
resource management, projecting resource utilization to optimize infrastructure usage.
• REQ-03: It is critical that the solution/system provides monitoring, dashboard and
dispatch capabilities for responsive observability, allowing operators to visualize and
interact with network configurations and projected traffic.
• REQ-04: It is critical that the solution/system includes security measures for continuous
monitoring of network traffic and automated enforcement of security policies.

43�4� Sequence diagram

172