AI-RAN Working Groups (WG)

As the telecommunications industry propels forward toward 6G, the integration of Artificial Intelligence (AI) into Radio Access Networks (RANs) is becoming not only beneficial but essential. To drive this transition, three specialized working groups have emerged

AI-for-RAN
AI-and-RAN
AI-on-RAN

Each targeting a distinct aspect of AI’s interaction with the RAN. These groups are actively shaping the future of telecom by pushing the boundaries of AI in network design, deployment, and operation.

AI-for-RAN Working Group

The AI-for-RAN Working Group is a collaborative initiative focused on integrating Artificial Intelligence (AI) into Radio Access Networks (RANs) to enhance their performance and efficiency. By embedding AI capabilities directly into network operations, the group aims to develop AI-native RANs capable of autonomous optimization, thereby improving overall network efficiency and user experience.

The primary objectives of the AI-for-RAN Working Group include:

Spectral Efficiency: Maximizing data transmission rates within the available bandwidth to accommodate increasing data demands.
Energy Efficiency: Reducing power consumption across network components to promote sustainable and cost-effective operations.
Processing Efficiency: Enhancing the computational performance of RAN systems to support complex operations and services.
Support System Efficiency: Streamlining operational processes through automation, reducing the need for manual intervention.

The AI-for-RAN Working Group engages in several key activities to achieve its objectives:

Literature Reviews: Conducting comprehensive analyses of existing AI and Machine Learning (ML) applications in telecommunications to identify best practices and areas for improvement.
Use Case Development: Identifying and defining scenarios where AI can effectively optimize RAN functions, leading to the development of targeted solutions.
Testing and Implementation: Conducting laboratory tests and implementing proof-of-concept systems to validate the effectiveness of AI-driven solutions in real-world settings.

Use Case Example

1. AI-Driven Interference Management

In densely populated urban environments, signal interference is a significant challenge that can degrade network performance. Traditional methods of interference management often struggle to adapt to the dynamic nature of such environments. By employing AI algorithms for real-time interference detection and mitigation, networks can dynamically adjust parameters to maintain optimal performance. This approach enhances spectral efficiency and ensures a consistent user experience.

2. Energy Savings through AI

AI can be leveraged to reduce energy consumption in RAN operations by analyzing network traffic patterns and dynamically adjusting the power states of network components. For instance, during periods of low traffic, AI algorithms can identify underutilized cells and temporarily deactivate them, leading to significant energy savings without compromising service quality.

3. Mobility Optimization

AI-driven mobility optimization involves using predictive analytics to manage handovers between cells more efficiently. By analysing user movement patterns and network conditions, AI can anticipate the need for handovers and execute them proactively, reducing dropped calls and improving connectivity. This enhances the overall user experience and network reliability.

Collaboration and Future Directions

The AI-for-RAN Working Group collaborates with industry leaders, academic institutions, and standardization bodies to ensure the seamless integration of AI into RAN systems. By aligning with existing standards and contributing to the development of new ones, the group aims to facilitate the widespread adoption of AI-driven solutions in the telecommunications industry.

Looking ahead, the group plans to explore advanced AI techniques, such as deep reinforcement learning and federated learning, to further enhance RAN performance. Additionally, the group is committed to addressing challenges related to data privacy, security, and the interpretability of AI models to build trust and ensure compliance with regulatory requirements.

By focusing on these areas, the AI-for-RAN Working Group aims to drive innovation and create intelligent, adaptive, and efficient RAN systems that meet the evolving demands of modern telecommunications networks.

AI-and-RAN Working Group

The AI-and-RAN Working Group is dedicated to exploring the concurrent utilization of converged compute-and-communications infrastructure to run both Radio Access Network (RAN) and Artificial Intelligence (AI) workloads. This integration aims to enhance platform utilization and unlock new monetization opportunities for network operators.

The group's primary objectives include:

Resource Optimization: Efficiently sharing resources between RAN and AI applications to maximize infrastructure utilization.
Multi-Tenancy: Supporting multiple applications on shared infrastructure, enabling diverse services to coexist without interference.
Service Quality and Security: Maintaining high standards of service quality and ensuring robust security measures while enabling resource sharing.

The AI-and-RAN Working Group engages in several key activities:

Solution Reviews: Evaluating existing technologies and methodologies for the concurrent operation of RAN and AI workloads.
Architecture Design: Developing frameworks and architectures that facilitate integrated RAN and AI operations.
Performance Evaluation: Assessing the effectiveness and efficiency of integrated systems to ensure they meet desired performance metrics.

Use Case Example

1. AI-Driven Network Optimization

By deploying AI applications within the RAN infrastructure, operators can implement real-time network optimization techniques. For instance, AI algorithms can analyze network traffic patterns and predict congestion, allowing for proactive adjustments to network parameters. This leads to improved user experiences and more efficient network operations.

2. AI-Powered Edge Computing Services

Integrating AI workloads at the network edge enables operators to offer services such as real-time video analytics and augmented reality applications. This approach reduces latency and enhances user experiences, demonstrating the potential of AI-and-RAN integration.

3. Dynamic Resource Allocation

The convergence of AI and RAN allows for dynamic resource allocation, where AI models predict and allocate resources based on real-time demand. This ensures optimal performance and efficient utilization of network resources.

4. AI-Enhanced Network Security

Deploying AI within RAN infrastructure can enhance network security by detecting and mitigating threats in real-time. AI models can analyze patterns and identify anomalies, providing a proactive approach to network security.

5. AI-Assisted Network Planning

AI can assist in network planning by predicting user behavior and network usage patterns, allowing operators to optimize network deployment and expansion strategies. This leads to cost savings and improved network performance.

AI-on-RAN Working Group

The AI-on-RAN Working Group, operating under the auspices of the AI-RAN Alliance, is dedicated to defining the radio interface requirements essential for deploying Artificial Intelligence (AI) and Generative AI (GenAI) applications across diverse sectors. By benchmarking these applications on current 5G networks, the group aims to identify enhancements necessary for the evolution toward 6G systems.

The group's primary objectives include:

Performance Benchmarking: Assessing the efficacy of AI applications operating over existing 5G networks to establish baseline performance metrics.
Requirement Identification: Determining the necessary advancements in radio interfaces to support seamless AI and GenAI application deployment in future 6G networks.
Security and Compliance: Ensuring that AI applications adhere to stringent regulatory standards, focusing on data privacy and network integrity.

The AI-on-RAN Working Group engages in several critical activities:

Technique Reviews: Evaluating current AI/ML and GenAI methodologies to understand their applicability and limitations within RAN environments.
Challenge Identification: Highlighting obstacles associated with deploying AI applications on RAN infrastructure, including latency issues, bandwidth constraints, and computational requirements.
Lab System Development: Creating controlled environments for rigorous performance testing of AI applications over RAN, facilitating the validation of theoretical models and practical implementations.

Use Case Example

Real-Time Language Translation Services

By integrating AI capabilities within the RAN, operators can offer real-time language translation services, enhancing cross-lingual communication for users. This application leverages the low-latency and high-bandwidth characteristics of modern networks to process and deliver translations instantaneously, significantly improving user experience in multicultural regions.

AI-Driven Multimedia Applications

The group explores AI-based multimedia applications, such as augmented reality (AR) and virtual reality (VR) streaming, which demand high data rates and low latency. Deploying these applications over RAN enables immersive experiences in gaming, virtual tourism, and remote collaboration, showcasing the potential of AI-on-RAN to revolutionize entertainment and professional interactions.

Industrial Automation and Robotics

In industrial settings, AI-on-RAN facilitates real-time control and monitoring of automated systems and robotics. For instance, deploying AI models at the network edge allows for immediate analysis and response to sensor data, optimizing manufacturing processes and enhancing safety protocols.

Healthcare Applications

AI-on-RAN supports telemedicine services by enabling real-time transmission and analysis of medical data. This capability allows healthcare professionals to perform remote diagnostics and consultations effectively, expanding access to medical services, especially in underserved areas.

The AI-on-RAN Working Group plays a pivotal role in integrating AI and GenAI applications within radio access networks, addressing technical challenges, and unlocking new service potentials across various industries. Through comprehensive benchmarking, requirement analysis, and the development of innovative use cases, the group contributes significantly to the advancement of intelligent, AI-driven network solutions poised to define the future of telecommunications.

The integration of Artificial Intelligence (AI) into Radio Access Networks (RAN) marks a fundamental shift in the evolution of wireless communications, particularly as we move toward 6G. The coordinated efforts of the AI-for-RAN, AI-and-RAN, and AI-on-RAN Working Groups under the AI-RAN Alliance reflect a comprehensive, multidimensional approach to embedding intelligence throughout the network stack—from RAN operation and resource management to application-layer service delivery.

The AI-for-RAN Working Group is laying the groundwork for AI-native RANs, where machine learning algorithms dynamically optimize network performance, improve energy and spectral efficiency, and adapt in real-time to changing traffic patterns and environmental conditions. Use cases like AI-based interference mitigation and automated network tuning illustrate the value of this direction.
The AI-and-RAN Working Group is enabling a converged compute-communications infrastructure, where both RAN and AI workloads co-exist on shared platforms. This allows for enhanced infrastructure utilization and monetization opportunities through multi-tenancy, especially in edge data centres. Real-world scenarios include low-latency AI inference at the edge for AR/VR applications and intelligent resource slicing in private networks.
The AI-on-RAN Working Group is pioneering the use of RAN as a delivery platform for AI and GenAI applications. By defining the radio requirements for hosting these services, the group ensures that future networks will be capable of supporting emerging AI-driven use cases such as real-time language translation, immersive multimedia, industrial robotics, and telehealth.

Together, these working groups form a unified vision for intelligent, autonomous, and service-aware networks. Their collective work is not only addressing current technical gaps but also shaping the architectural, operational, and regulatory foundations of future 6G systems. By aligning industry stakeholders, academic researchers, and standards bodies, the AI-RAN Alliance is accelerating innovation at the intersection of AI and wireless communications—paving the way for a truly intelligent network infrastructure that meets the complex needs of future digital societies.

References

AI-RAN Alliance Whitepaper, AI-RAN.org
3GPP Technical Reports on AI in RAN, 3GPP.org
O-RAN Alliance Specifications, O-RAN.org
SoftBank & NVIDIA Collaboration on AI-RAN, NVIDIA.com
Ericsson Research on AI for RAN Optimization, Ericsson.com
Nokia AI-Powered Networks Whitepaper, Nokia.com
T-Mobile USA AI-RAN Deployment Strategy, T-Mobile.com
IEEE Transactions on Wireless Communications: AI and RAN Integration
AI in Telecommunications: A Future Perspective, ITU Reports
SoftBank's AITRAS System for AI-RAN, SoftBank.jp
SoftBank, AI-RAN, and D-MIMO (Analyst Angle)
AI-RAN Goes Live and Unlocks a New AI Opportunity for Telcos | NVIDIA Technical Blog
SoftBank releases edge AI-RAN solution to transform telecom by 2026 | Edge Industry Review
SoftBank, OpenAI set up AI joint venture in Japan
SoftBank, Ericsson enhance AI-RAN collaboration | Computer Weekly
Press Release on AI-RAN Development | About Us | SoftBank
SoftBank Unveils AITRAS Converged AI-RAN Solution - SDxCentral
AI-RAN: The Social Infrastructure Supporting the AI Era | SoftBank Research Institute
SoftBank Shares Hit Record High Amid Push Into AI And Computer Chips
SoftBank Corp. Announces Development of “AITRAS,” a Converged AI-RAN Solution | Business Wire

AI-RAN Working Groups (WG)

Recent Posts

Comments