Realize "ease of connection" not only during normal times but also in emergencies and event times, supporting cost reduction and power saving.

Fujitsu Limited, as part of NEDO's commissioned project "Post-5G Information and Communication System Infrastructure Strengthening Research and Development Project" (hereinafter referred to as the project), has developed an application that advances network operations by utilizing artificial intelligence (AI) technology to enhance communication quality of mobile networks while achieving power saving, including world-first technologies.

Fujitsu has a track record of adoption in the global market for radio units (RU), and in the future, it plans to globally provide applications developed by the company, which harness the footprint cultivated in RUs, starting from November 2024 to mobile network operators worldwide, based on the operation management system (SMO) "FUJITSU Network Virtuora Service Management and Orchestration" according to the O-RAN specification.

This application consists of three technologies: estimating network quality in real-time with AI to maintain quality, technology to prevent degradation of network quality in situations like event times in advance, and technology to redesign coverage areas of base stations to maintain quality. The effectiveness has been confirmed in verification under conditions close to operational environment.

This enables mobile network users to achieve the most anticipated "ease of connection" not only during normal times but also in emergencies such as natural disasters or during events, leading to improved convenience, satisfaction, and safety in emergencies. For mobile network operators, it supports cost reduction and power saving through appropriate operations according to traffic volume, contributing to solving global social issues.

1. Background

Amid rapid global digital transformation (DX), the proliferation of 5G mobile networks as one of its infrastructures is expanding. Mobile network operators that support this are expected to further enhance functions such as ultra-low latency, simultaneous connections, and ensure network quality on a per-user and per-application basis in the future. In the radio access network (RAN) domain, open and virtualization based on the O-RAN concept are progressing, with expectations for TCO (Total Cost of Ownership) reduction.

Based on these backgrounds, Fujitsu, in NEDO's implementation of this project, developed three applications to improve sensory quality (QoE), achieve power saving, and maintain communication quality, which operate on the RIC (Reference Implementation Configuration) conforming to O-RAN and placed inside the SMO (Network Service Operations Management System) supporting intelligence like RAN autonomization and automation. The effectiveness has been confirmed through technical verification utilizing commercial data from mobile network operators, and the network operations management software 'Virtuora' series will start providing from November 2024 onwards.

2. Results of this time

(1) Utilizing AI to achieve real-time estimation of QoE of mobile network users and quality assurance

We have developed a technology that estimates QoE in real-time and automatically switches to the network area of other base stations when a deterioration in QoE is detected.

This technology is the world's first to generate an AI model that estimates the QoE of each application easily by selecting feature quantities from statistical data (KPI) calculated at the user and application levels through high-speed packet analysis capable of handling 100Gbps RAN traffic, allowing flexible adaptation to a wide range of applications.

As a result, by accurately grasping the QoE of each user and allocating necessary resources, it is possible to improve the number of accommodating users per base station by 19% (Note 7) by ensuring convenience and satisfaction while suppressing excess resources.

Figure 1: QoE estimation based on packet analysis

(2) Detecting premonitory signs of increased communication traffic and achieving quality maintenance and power saving through the start and stop of base stations

In case of emergencies like natural disasters or event hosting, AI detects the increase in communication traffic from normal times, allowing base stations that were previously in sleep mode to be activated in advance and prevent deterioration in communication quality for users.

Previously, we focused on monitoring traffic in each area in real time and achieving power savings by putting base stations that do not need to be activated to sleep. In addition to this, we have developed a world-first technology that detects abnormal increases in traffic at the grid unit level following events like local events with increased foot traffic compared to normal times. With this technology for early detection, we were able to activate base stations in advance without impacting user quality for 99.8% of the validation period.

Based on this, performing detailed base station activation and deactivation according to traffic conditions, combined with the low power consumption application announced by Fujitsu in December 2023, enables both QoE maintenance and power saving. This contributes to improved convenience, satisfaction, ensuring safety during emergencies, and addressing social issues.

(3) Detection of deteriorating service quality and maintenance through area redesign

In abnormal detection technology for a single cell, it was difficult to determine whether the cause of traffic decrease was a simple load decrease or an anomaly. With this technology, instead of focusing on a single cell, comparing with surrounding cells and traffic trends allows AI to make judgments, achieving high fault detection accuracy (precision above 92%). It also supports supervised learning with limited fault data and unsupervised learning. Moreover, understanding the service impact considering cell overlay situation enables prioritizing areas for recovery.

For areas determined to have a significant impact on services by this abnormal detection technology, this technology further calculates the optimal tilt angle in the surrounding cells by considering the direction and load conditions of neighboring cells and a radio wave propagation prediction model taking into account real field path loss, minimizing the impact on service quality due to a faulty cell. This successfully reduces the recovery time from about a day to less than an hour during abnormal occurrences like equipment failures, minimizing the impact on users.

Figure 2: Detection and recovery of service quality degradation

3. Future Plans

Fujitsu will continue to contribute to the realization of a more secure, safe, and sustainable society by providing O-RAN products utilizing AI-centered technology to support networks for various industries.

NEDO aims to strengthen the development and manufacturing infrastructure of Japan's post-5G information communication system by developing core technologies that will be compatible with post-5G information communication systems in the future, starting with this technology.

Trademarks

Proper nouns such as product names mentioned are trademarks or registered trademarks of their respective companies.

Notes

• Note 1
  O-RAN:
  This refers to RAN (Radio Access Network) built based on the open interface specification set by the O-RAN ALLIANCE.
• Note 2
  This project:
  Project name: Post-5G information and communication system infrastructure strengthening research and development project / Development of Post-5G information and communication system / Development of RAN control advanced technology (commissioned)
  Project duration: From fiscal year 2021 to fiscal year 2024
  Project overview: Post-5G information and communication system infrastructure strengthening research and development project
• which utilizes generated AI, members who newly develop a manufacturing process for laundry detergent for clothing can easily search and utilize skilled workers' technology, knowledge, and know-how and, thus, support the efficient execution of the business task.
  SMO (Network Service Operation and Management System):
  In the O-RAN architecture, it refers to systems that optimize the operation and maintenance of RAN defined by Service Management and Orchestration (SMO), managing the lifecycle as well.
• Note 4
  RIC:
  RAN Intelligent Controller (RIC) is an abbreviation for a controller that autonomously executes and controls the optimization of RAN parameter settings and operations.
• Note 5
  QoE (Quality of Experience):
  Quality of Experience (QoE) refers to the satisfaction and comfort felt by users when using services such as video streaming, online games, applications, etc. Various elements like image quality and responsiveness affect QoE.
• Note 6
  Commercial data of mobile network operators:
  The data collected by mobile network operators in actual operations is referred to be utilized for AI development in this research and development.
• Note 7
  19% improvement:
  Calculated from the results of comparing cell-based performance measurement and QoE-based performance measurement in Fujitsu's lab environment.
• Note 8
  Grid:
  Refers to dividing the latitude and longitude on a map into grid sections.
• Note 9
  Power-saving application announced in December 2023:
  This refers to Fujitsu's power-saving application that uses AI technology to estimate communication traffic based on the distribution of location information of mobile network users.
  (Reference) Fujitsu Press Release (December 19, 2023) 'Achieving power saving in network operations through the use of AI'
• Note 10
  Cell:
  It refers to the minimum unit of a communication area formed by base station radio waves.
• Note 11
  Pass loss:
  It refers to the attenuation received when radio waves pass through the air or are obstructed by obstacles.

4. Inquiry regarding this matter

(Contact for inquiries regarding the contents of this news release)

NEDO

Fujitsu

(General inquiries regarding other NEDO projects)

The price, specifications, service contents, etc. of the product described in the press release are as of the announcement date, and may be changed without notice. Thank you for your understanding.