Our blog series on the journey to autonomous networks has tracked the path to AI-based autonomous networks and the challenges of building those networks.
With some of the important strategy discussion in the rearview, this post takes a more in-depth look under the hood at actual automation use cases telcos can deploy. Our work within Rakuten has revealed that use cases powered by Level 3 automation can begin to have significant impact on the business.
Level 3 automation, known as conditional automation, marks a significant advancement in network management. At this stage, AI begins to autonomously generate insights, although it does not execute actions autonomously based on these insights. The automation of actions is reserved for higher levels (4 and 5), where AI not only suggests but also implements solutions autonomously.
Before we delve into the use cases, it's important to note that while discussing various levels of AI automation and autonomy is insightful, the ultimate goal is not merely to achieve these levels. Instead, our focus should be on addressing the specific challenges your network faces.
In the journey to an autonomous network, sometimes the path will mean moving vertically to new levels and other times it will be about moving horizontally to bring AI deeper into a particular process. This might also be different in different network domains.
While having a taxonomy makes AI autonomy roadmap easier to understand, I don’t want the focus on solving the most pressing network problems to get lost in the discussion.
With this in mind, here are four Level 3 applications that can add immediate impact for a CSP:
Typical ATP activities include site surveys, installation and drive tests, each of which are completely manual processes for many CSPs. Because they are manual, every step in each of these procedures can be subject to delay or error. AI-based automation can develop insights into the way the network sites are built and validated, thus streamlining processes and making them more deterministic. AI-based image recognition can assist in auto-approving the site audits and hazard checks. AI can perform auto-correction for a failed drive test report to meet the acceptance parameters.
Every telecom system has a management dashboard where data is displayed to be measured against the key performance indicators (KPIs). AI automation can turn this data into insights to be acted upon by network managers. This is a big step up from network managers needing to process the data and sets the stage for further automation that can act on the insights. Instead of just being information provider tools, AI can extract insights out of the dashboards and reports.
Building out a network site involves many processes. Some of them are on the critical path and must be done as efficiently as possible. Others are costly or are complex. AI can be used to simplify and optimize these processes, reducing the time needed to complete these processes.
FCAPS is the industry standard paradigm for network management incorporating fault, configuration, accounting, performance and security management levels. With AI, FCAPS management can move to autonomous monitoring whereby the system is automatically detecting anomalies and reporting them instead of network managers creating the rules based on their past experiences.
These use cases provide a sample of how AI automation can move a network down the path to autonomy and solve key network challenges.
My next post will explore Level 4 AI use cases that introduce root cause analysis and use AI to suggest next best actions.
