AI-based automation becomes powerful when levels four and five of the network AI autonomy scale are reached (see below for a refresher on autonomy levels).
These levels support network systems controlled by AI-infused network and service management, collecting and logging network statistics, reporting them to managers and fixing problems via instantaneous, automated network parameter or network function adjustments.
Our previous posts explored the six-level path to AI-based autonomous networks, challenges of building those networks and how level-three use cases can impact the network.
According to the automation taxonomy, Level 4 offers AI-assisted automation and Level 5 offers complete intent-aware autonomy. Often, the difference between these levels is the amount of human intervention required.
A fully autonomous network comprises all Level 5 systems, but in reality, MNOs seek to automate the systems that create the biggest pain points. That’s why thinking about the path to network autonomy as a series of use cases is beneficial.
Let’s explore three Level 4 use cases that will have a dramatic impact on an MNO network:
RCA and NBA are the found data of service assurance. Adding AI makes it possible for these techniques to keep up with new network dynamics like continuous deployment and redeployment of cloud native clusters, including software micro services. As we have discussed in our Zero Touch Telecom newsletter, having an enterprise data model and right data strategy is important to realize this level of autonomy. The AI functionality enables the system to know who is doing what and when, based on continuous monitoring and action orchestration and then to recommend RCA and NBA autonomously.
Collecting traces of signaling data flows provides a tried-and-true diagnostic data source for common network configuration or interoperability problems. These are issues that reduce coverage, capacity or both. This Level 4 automation capability uses the trace data for autonomous capacity and coverage optimization (CCO) via an algorithm or an rApp for a RAN intelligent controller. We have implemented many use cases where AI assistance was used for some of the most common problems like sleeping cell detection and remediation as well as new areas of innovation around energy management.
An observability dashboard serves as the window into every event in a Kubernetes cluster. Therefore, organizations must build custom dashboards that offer a view of health, capacity, software changes, audits and insightful analytics into the Kubernetes cluster. With this Level Four tool, data from observability and policy systems are gathered, and suggested changes are presented for action. Once a network engineer makes the changes, this system pushes the changes to the network elements for implementation. This allows MNOs to shift focus from dashboards to insights. We simply cannot have more dashboards for increasingly complex networks. AI-based logs and monitoring is the only sustainable way to manage the increased monitoring-data overload.
Let’s move on to two Level 5 use cases that can provide full autonomy to an MNO network:
Site reliability engineers (SREs) must meet service level-agreements (SLAs) and key performance indicators (KPIs) when working to restore a service outage. With a large language model (LLM)-based incident management and restoration system, SREs can create scripts that meet the goal of automating the response to future-like situations in a way that reduces the time it takes to restore the network. With the aid of generative AI, network operations engineers can immediately do the work of coders and play an active role in automating key processes.
Using an AI co-pilot assistant to manage network incidents can significantly reduce the mean time it takes to restore the network after an incident. Using AI, operations support engineers or incident managers can streamline ITSM tasks, improve accuracy and accelerate issue resolution. A level five AI-based incident copilot can manage the trouble ticket triaging, reassignments, execution of critical restoration activities, tracking the resolution, summarization and communication related to the incident, root cause analysis review and incident debrief including lessons learnt.
As these use cases demonstrate, reaching Level 4 and Level 5 automation can have a dramatic impact on network operations and service quality.
We are constantly working on new use cases using our AI technology and the experience we have from Rakuten Mobile. The job of network autonomy is never done. But by applying AI automation to problem areas of your network you can make a big difference.
