NTT has been conducting research and development in areas such as communication quality, traffic control, and network operations that utilize AI. We believe that supply and demand visualization and matching will become more important factors in the ICT infrastructure supporting a variety of social services; we call these technologies "MoHFI" (Matching of Human Feeling/Intelligence) as we research and develop them.
Here we will be discussing self-evolving zero-touch operations, which offer a service while autonomously adapting to changing demand and network conditions, as well as user-engagement operations, which provide a more agreeable service for users while reducing network and service costs. We will also introduce multiplayer collaboration and value-added traffic data, which we have recently embarked on.

Achieving Self-Evolving Zero-Touch Operations

NTT has been conducting many different kinds of research with the goal of automating and optimizing network operations, including technologies that promptly detect changes in network conditions and failure, technologies that locate and identify the causes of failure, and traffic-classification and prediction technologies that automatically classify traffic with similar characteristics.
Based on these technologies, we are now aiming to create self-evolving zero-touch operations, where the network itself is capable of independently adapting and learning from changes in demand or environment through the use of AI, as well as self-evolving upon encountering unknown situations such as social changes.

A roadmap for self-evolving zero-touch operations is shown in Figure 2. We are currently at Level 1, where work is automated by humans. Going forward, we aim to move on to Level 2, where operations are autonomously optimized through human-AI interaction, then Level 3, where the AI is able to autonomously optimize operations of unknown works, and finally Level 4, where the AI can autonomously optimize all work. Level 4 will also require processing of work that is difficult to anticipate in advance.

NTT Network Technology Laboratories, NTT Network Service Systems Laboratories and NTT Access Network Service Systems Laboratories are working together to develop the technology required to reach Level 4. NTT Network Technology Laboratories is focusing its efforts on the following 4 areas:
(1) AI Advancements
Technology that aims to autonomously follow changes occurring inside and outside the network, such as an increase or decrease in equipment and changes in the way users utilize the service, as well as expanding the data used for analysis.
(2) Visualization of AI Decision Logic
Technology that allows humans to follow the rationale and thought process the AI takes to optimize processes in light of changes both inside and outside the network by visualizing it.
(3) AI Quality Management
Technology that audits the level at which changes inside and outside the network are detected and to what extent the detected changes can be dealt with.
(4) Evolution of AI by Learning in a Pseudo-Environment
Technology where digital twins in cyberspace simulate a variety of actions, which is used to evolve the AI through learning, making it capable of reacting to unknown faults or disasters.

Enabling User-Engagement Operations

User-engagement operations refers to technology that provides an agreeable service for users while reducing the cost to the company of providing the network or service. It estimates the user's demands and willingness to continue using the network or service based on their activity, and encourages continued use without the user's conscious awareness.
In order to implement user-engagement operations, we need four peripheral technologies: technology to measure and visualize the service's user experience quality; technology to index and quantify the performance and quality of devices and networks; technology to analyze and predict future service usage through observed data; and technology to control and optimize the service by making assumptions based on the outcome of analyses and predictions (Figure 3). In the future, we will aim to combine these four technologies to deliver services with high user engagement based on the demands of service providers and end-users for a variety of resources, including wired/wireless networks, servers and devices.

Toward Multiplayer Collaboration

Modern network services consist of an interconnected structure of network operators, cloud operators and service providers. This makes it difficult to determine the system status of other operators, and creates challenges for end-to-end quality design, monitoring and control.
Until now we have mainly focused on self-evolving zero-touch operations and user-engagement operations in relation to our own field, but in order to solve these challenges, we plan to carry out research into situational predictions and resource control as a means of collaborative control alongside other network operators and service providers in other domains, such as energy.
Our aim is to establish an ICT infrastructure in the IOWN era that supports a variety of social services by collaboratively controlling multiple resources with differing compatible control devices, control policies, etc.

Toward Value-Added Traffic Data

Detection and optimal control of network errors requires the use of various different types of data, including data related to communication traffic and quality, data relating to network and system logs, performance data, and external data from sources such as Twitter.
The process from acquiring this data to converting it to a general-purpose format requires separate processing based on the data involved, but it can then be used for general processing for analysis or characteristic data extraction. We aim to create new value-added traffic data solutions where multiple processes are standardized, combined and offered together based on purpose.

NTT will continue to promote research and development in the four areas of self-evolving zero-touch operations, user-engagement operations, multiplayer collaboration and value-added traffic data, with the goal of achieving fully autonomous network operation.

*The names of the laboratories mentioned in the article may have changed since the time of writing/interview.