Press Releases

March 3, 2026

NTT and DOCOMO Demonstrate the World's First Technology to Estimate in Advance Whether Each Network Slice Can Meet Communication Requirements for Stable Communications
— Providing visibility into each slice's suitability for meeting communication requirements and contributing to improved stability and more advanced operational planning for social infrastructure and industrial applications —

• 
• Tweet

NTT, Inc.
NTT DOCOMO, INC.

News Highlights:

1. Developed the world's first^*1 technology that enables advance estimation, on a per-slice basis, of whether communication requirements for stable communications can be met by estimating and adjusting intermediate indicators related to communication characteristics in network slicing^*2 environments.
2. Verified the technology on the commercial network of DOCOMO and confirmed that each slice's suitability for meeting communication requirements can be assessed in advance for slices with different use cases.
3. This technology is expected to contribute to improved operational planning and risk reduction on an area basis for social infrastructure and industrial applications, including information sharing during disaster response, remote monitoring of social infrastructure, and inspections using drones.

TOKYO, JAPAN, March 3, 2026 --- NTT, Inc. (Headquarters: Chiyoda-ku, Tokyo; President and CEO: Akira Shimada; hereinafter “NTT”) and NTT DOCOMO, INC. (Headquarters: Chiyoda-ku, Tokyo; President and CEO: Yoshiaki Maeda; hereinafter “DOCOMO”) have successfully demonstrated a technology that enables advance estimation of whether communication requirements necessary to support stable communications can be met for each slice with different use cases.

The technology enables per-slice estimation, which had previously been difficult to achieve, by estimating and adjusting intermediate indicators related to communication characteristics. Its effectiveness was confirmed through verification experiments using commercial network equipment.

Going forward, the companies will further explore mechanisms based on this technology to enable advance assessment of each slice's suitability for meeting communication requirements across multiple performance metrics.

By enabling advance visibility into whether communication requirements can be met for each slice, the technology is expected to contribute to improved service stability and more efficient operations in social infrastructure and industrial sectors.

Background

With the growing adoption of cloud services, AI, and IoT, the need to handle large volumes of data and perform real-time processing is becoming increasingly important in social infrastructure and industrial applications. Services such as drone operations, robot control, and remote monitoring depend on stable mobile communications for efficient operation. However, in areas where traffic tends to concentrate, such as urban centers and tourist destinations, network congestion can lead to unstable communications, which may disrupt service delivery.

To address these challenges, DOCOMO is working to leverage slicing, a communication stabilization technology supported by its 5G SA^*3 network. Network slicing enables multiple services with different requirements to be handled simultaneously on the same network while providing dedicated network slices tailored to specific use cases.

For slicing to be effectively used in social infrastructure and industrial applications, it is important for service providers to understand in advance where and to what extent communication requirements^*4 can be met. For example, during disasters, temporary response bases or evacuation areas may experience concentrated communication demand within a limited area, which can make the sharing of video and sensor data or communication among relevant organizations unstable. If locations within an area where communication requirements are more likely to be met can be identified in advance, service providers can plan the placement of operational bases, communication routes, and operational frameworks in advance, including methods for video sharing and terminal deployment. This can help reduce operational risks and minimize the need for adjustments in the field. Improving the accuracy of such advance planning can also support faster initial response and reduce the burden on personnel involved.

In this demonstration, the companies confirmed that it is possible to estimate, on a per-slice basis, the extent to which communication requirements can be met within a given area, providing a technological approach for advance assessment.

Research Results

To enable the reliable use of network slicing in social infrastructure and industrial services, it is important to establish a mechanism that allows service providers to understand in advance, on a per-slice basis, each area's suitability for meeting communication requirements for expected use cases. Achieving this requires a technology capable of estimating whether communication requirements, such as throughput, can be met for each slice actually provided on the network.

However, in a network slicing environment, multiple slices operate while sharing the same network infrastructure. As a result, resource utilization in one slice can affect others. Conventional approaches have been limited to simplified estimates, such as applying a fixed coefficient to the overall effective network throughput, making it difficult to accurately estimate the extent to which communication requirements can be met on a per-slice basis.

To address this challenge, NTT and DOCOMO have newly established a technology that estimates each slice's suitability for meeting communication requirements by incorporating per-slice communication stabilization information into a machine learning–based estimation process. With this technology, even in environments where multiple slices are used simultaneously, it becomes possible to estimate more accurately how well each slice can support specific use cases based on indicators such as throughput.

Technical Highlights
Estimating intermediate indicators that influence communication characteristics to assess each slice's suitability for meeting communication requirements

In network slicing, different communication characteristics can be configured depending on the intended use case. To reflect these configurations when estimating whether communication requirements can be met, it is important to appropriately combine information on infrastructure-side resource allocation with machine learning–based estimation.

Rather than directly estimating suitability for meeting communication requirements based on infrastructure data, this technology first uses machine learning to estimate intermediate indicators that influence communication characteristics, such as radio quality metrics, the number of multiplexed radio connections, and the amount of resources used for communication stabilization. The stabilization policies configured for each slice are then reflected in and used to adjust these intermediate indicators. The adjusted values are subsequently used to estimate whether each slice can meet communication requirements based on indicators such as throughput.

Specifically, various types of data collected from network infrastructure are first used as inputs to estimate intermediate indicators using machine learning. Next, if the estimated values conflict with the communication characteristic policies configured for a slice, the values are adjusted to align with those policies. Finally, the infrastructure data together with the adjusted intermediate indicators are used as inputs to estimate how well each slice can support its intended use case (for example, in terms of throughput based suitability for meeting communication requirements) (Figure 1).

This approach makes it possible to estimate, in a manner that reflects actual operating conditions, each slice's suitability for meeting communication requirements while incorporating the communication characteristics configured for that slice.

Demonstration Results

To verify the effectiveness of the technology, NTT and DOCOMO conducted demonstration experiments in December 2025 using commercial network infrastructure in Tokyo. In addition to on-site measurements using actual devices, the companies performed estimations using data collected and accumulated from base stations on DOCOMO's commercial network and evaluated the difference between measured and estimated values.

1. Confirmation that per-slice estimation is consistent with measured values (RMSE reduced by 51% in a reference comparison over a specific period)
   

   First, the companies evaluated whether communication requirements could be estimated on a per-slice basis by estimating terminal throughput on operational infrastructure. Conventionally, approximate values have been calculated by applying a fixed coefficient to the overall effective network throughput. However, this approach does not sufficiently reflect the impact of resource utilization across slices.

   In this demonstration, the estimation results were validated by comparing them with measured values using continuously accumulated data from commercial network infrastructure. The results showed that terminal throughput estimated using the new technology achieved improved accuracy compared with measured values over the entire observation period. In addition, in a reference comparison over a specific period using the conventional approximation method as a baseline, the error (RMSE)^*5 was reduced by 51%, demonstrating that each slice's for specific use cases can be estimated more accurately under realistic conditions.

2. Visualization of estimation results using heat maps to identify locations within an area where communication requirements are more likely to be met
   

   Using data collected from base stations on DOCOMO's commercial network, the likelihood of meeting communication requirements was estimated based on terminal throughput and visualized in the form of heat maps (Figure 2). The results showed that this technology enables visual identification of locations within an area where communication conditions are more suitable for specific use cases.

   

   These results confirm that by analyzing data accumulated on commercial networks, the proposed technology can accurately estimate the extent to which communication requirements will be met for each slice and visualize differences across locations within an area. Based on this information, service providers can incorporate communication feasibility into the planning stage when determining base station placement, communication routes, and device deployment.

   For example, during disaster response operations, the technology can help identify in advance where to place terminals used to share video, maps, and sensor data around temporary response bases or evacuation areas, and which communication routes are likely to be more stable. This helps reduce the need to relocate equipment or reconfigure systems after arriving on-site due to unstable communications.

   By enabling service providers to utilize slices with a clearer understanding of whether required communication performance can be achieved, the technology is expected to support the effective use of network slicing in 5G SA environments and contribute to the broader adoption and deployment of slicing-based services.

Roles of Each Company

• NTT
  Development of a technology that enables advance estimation of whether communication requirements necessary to support stable communications can be met, taking into account per-slice communication stabilization control.
• DOCOMO
  Provision of the demonstration environment and accumulated network data from the commercial network, evaluation of the demonstration results, and exploration of services utilizing the proposed technology.

Future Outlook

Going forward, NTT and DOCOMO will build on the technology demonstrated in this study and further develop the estimation and visualization framework. In addition to throughput, the companies will explore mechanisms that enable advance assessment of each slice's suitability for meeting communication requirements across multiple performance metrics, including latency.

By enabling advance visibility into whether communication requirements can be met for each slice, the technology is expected to contribute to improved service stability and more efficient operations in environments where communication conditions may fluctuate, such as information sharing during disaster response, remote monitoring of social infrastructure, and inspections using drones.

Through the continued evolution of slices designed to support diverse use cases, NTT and DOCOMO will work to address social challenges and create new value through network slicing technologies.

【Glossary】

1. Source: NTT, as of March 3, 2026.
2. Network slicing
   A technology in 5G SA that virtually partitions (slices) a network to provide networks tailored to a wide range of requirements and use cases.
3. 5G SA
   A communication architecture that combines a 5G-specific core network infrastructure, known as 5GC (5G Core), with 5G base stations.
4. Communication requirements
   Performance conditions required for the stable operation of a service. Examples include maintaining terminal throughput above a certain level and keeping latency below a specified threshold. In this demonstration, throughput was used as the evaluation indicator.
5. RMSE (Root Mean Squared Error)
   A metric used to measure the magnitude of estimation error. It represents the square root of the average of the squared differences between estimated and measured values.