Successful Verification Experiment of Network Operation Using Dialogue with AI

In an age where everything is connected to communications, networks have become an essential part of the social infrastructure, and there is a demand for them to maintain a high level of quality. To maintain the quality of its network, KDDI is promoting the automation of monitoring operations as part of its "Smart Operation Project" [1], and in 2024 it began operating an AI-based fault detection system for its LTE mobile network [2].
On the other hand, the reality is that much of the knowledge required to build and maintain networks depends on operators with many years of experience and expertise. In addition, as networks become more complex and the number of countermeasures for events involving large gatherings increases, the limitations of manual responses-such as configuring individual settings on each network device-are becoming apparent.
To address this situation, both companies explored the use of AI as a way to autonomously build, configure and manage networks with optimized device settings and operational policies.

Using this system, the two companies successfully verified the autonomous configuration and management of a network that satisfies the operator's requirements, such as "enhancing the network for the music festival" and "improving traffic imbalance." The system was also confirmed to reduce the operators' workload and prevent human error by operating autonomously.
For example, when implementing countermeasures for large events such as music festivals, a natural language exchange with the AI provides details about the expected traffic volume and other data and creates a Network Intent that describes the requirements for the network. By translating this into settings for each device, the controller on Kubernetes can meet network requirements, allowing configuration changes up to the network level while reducing operator workload and preventing human error.