Starting Operation of "Fault Recovery support Agent" Which Identifies Cause of Failure Immediately

―Enables automatic identification of cause of failure in voice calls or data communication in cloud environments―

KDDI Corporation

Tokyo, February 19, 2026―KDDI started operation of the Fault Recovery support Agent (hereinafter referred to as the AI Agent) which immediately identifies the cause of a failure in some services built in cloud environments including voice calls, data communication, and au PAY on February 19, 2026.

The AI Agent identifies the cause of a failure by comprehensively analyzing correlations between affected services and systems, equipment alarms that were issued, maintenance operations that were conducted, etc. Structurization of the information of each service's networks and systems enables the AI Agent to analyze it. This allows operation staff to easily identify the cause of a complicated failure that affects multiple systems.

In addition, KDDI plans to introduce the "Autonomous Maintenance Agent" in FY2026. The Autonomous Maintenance Agent will collaborate with the Fault Recovery support Agent to carry out specific recovery measures such as isolation of equipment and conduct maintenance operations necessary for replacing failed parts, etc. By utilizing the AI Agents, KDDI aims for operational DX that automates everything from identifying the cause of a failure to carrying out recovery measures.

KDDI will continue making efforts at operational DX utilizing various technologies including AI and aiming to provide customers with high-quality and highly-stable telecommunications services.

■Background

When a failure occurs in a voice call, data communication, or service, initial response measures including identification of the cause of the failure and affected range are carried out, and then a specific recovery operation is started.
Since 2021, KDDI has devoted efforts to "Smart Operation [1]" which comprehensively monitors various services and carries out a recovery operation through a one-touch operation or, if possible, zero-touch operation. It enables swift recovery of services from a failure in a single system.
On the other hand, when a complicated failure occurs involving multiple systems, it still takes time and effort to identify the cause of the failure and recover the equipment status to that before the failure, requiring further evolution of "Smart Operation." The AI Agent enables immediate identification of the cause of a failure, which conventionally takes time during the initial response.

■About AI Agent

KDDI builds a digital twin for operation by structuralizing the information of services and system configurations.
When the AI Agent detects a failure involving multiple systems through alarms, it analyzes correlations among services and systems in which the failure occurred by using a centrality analysis [2] on the digital twin for operation. After that, from among a lot of alarms, it immediately identifies the system that is likely to be the cause of the failure.
Furthermore, it increases the accuracy of the estimation of the cause of the failure by combining multiple pieces of information including the equipment alarms that were issued and maintenance operations that were conducted. The AI Agent presents the identified cause to the operation staff, supporting the recovery operation.

<Image of service correlation analysis on digital twin for operation>

■About exhibition at "MWC26 Barcelona"

KDDI will present an exhibition of network operations utilizing AI at the world's largest mobile industry trade show, "MWC26 Barcelona," which will be held in Barcelona, Spain, from March 2 to 5, 2026.

"MWC26 Barcelona" special website

[1]
- News Release on July 15, 2021
  A new operation base was launched to operate integrated KDDI services and provide more stable telecommunications services (in Japanese only)
- News Release on November 12, 2021
  Revamped Osaka operation base through automation to reduce restoration time by up to 40% compared with the previous system (in Japanese only)
[2]A technique used in graph theory. It is used to quantitatively analyze the positional ranking of each node (vertex) in a network. The Fault Recovery support Agent uses this technique to carry out an analysis focusing on the role of the function provided by each system in a telecommunications service.

*The information contained in the articles is current at the time of publication.
Products, service fees, service content and specifications, contact information, and other details are subject to change without notice.

Downloads

Download Images