The evolution of RAN and the role of AI in network optimization

How can Fujitsu use AI to improve network optimization?

In our fast-paced, interconnected world, wireless networks are at the heart of everything we do. From video sharing and online gaming to remote work and industrial automation, the demand for reliable and efficient network connectivity continues to grow. The advent of 5G has significantly improved network capacity, latency, and the ability to support new types of services and end points. As we look to the emergence of 6G, momentum is building for even more advanced features and environmental considerations, particularly power consumption, to enable better overall network optimization.

One of the critical components of any wireless network is the Radio Access Network (RAN), which connects various devices like smartphones, vehicles, industrial machinery, and even virtual reality devices. The orchestration and operational sophistication of RANs are becoming increasingly important as they play a vital role in ensuring seamless network performance.

This blog post explores the evolution of RAN and its integration with Artificial Intelligence (AI) for enhanced network optimization and management. We will also examine the standardization efforts by organizations like the 3^rd Generation Partnership Project (3GPP) and the O-RAN Alliance, as well as the advanced AI technologies developed by Fujitsu to address the challenges faced by RANs.

Virtualization and other network trends

Traditionally, network components like base stations, switches, and routers were hardware-based. However, the rise of virtualization has enabled the decomposition of these components, making it possible to configure flexible network setups by combining elements from multiple vendors. Intelligent orchestration, which reassembles distributed components effectively, is crucial for creating energy-efficient networks. RANs play a vital role in connecting a wide array of devices, so their orchestration and operational sophistication are essential points of network control.

Standardization in RAN

Standardization is pivotal in ensuring interoperability and the seamless evolution of wireless communication systems. Two significant organizations leading the charge in RAN standardization are the 3GPP and the O-RAN Alliance. The 3GPP organization is focused on standardizing mobile communication systems and is actively developing 5G-Advanced, an evolution of 5G. This evolution involves incorporating intelligent orchestration, AI, and Network Data Analytics Function (NWDAF) to enhance network efficiency and energy conservation.

The O-RAN Alliance, on the other hand, has two visions: open and intelligent. Their architecture aims to achieve these visions by utilizing platforms like the RAN Intelligent Controller (RIC), which optimizes RAN resource management and automates operations. Depending on the control cycle, this controller further divides into Non-Real Time RIC (Non-RT RIC) and Near-Real Time RIC (Near-RT RIC). The flexibility of open interfaces like A1, E2, and O1 allows for multivendor systems with elements from different manufacturers.

The O-RAN Alliance has been working on O-RAN compliant software, implementing the AI/ML (Artificial Intelligence/Machine Learning) framework, which was released in December 2022.

The future of network technology

Wireless access technology, including base station equipment, will continue to evolve, focusing on performance improvements and multivendor integration. Performance enhancements will include faster communication speeds and lower latency, facilitated by high-frequency communication technology and edge computing. Integrations will involve virtualized network functions and support for open interfaces standardized by the O-RAN Alliance.

New network operation technologies will also emerge to monitor and control RANs, providing network slicing, optimal resource management, and energy efficiency. Intelligent orchestration will be essential in delivering flexible control to the entire RAN.

Challenges in RAN operations

The increasing complexity of networks due to service diversification and growing power consumption are significant challenges in RAN operations. Networks face continuous fluctuations in connections and quality due to short-term changes in user behavior, events, or physical communication environments. Whether responding to sudden disasters or meeting Service Level Agreements (SLAs), the challenge lies in quickly and effectively combining elements to support diverse service delivery requirements.

With commitments to reduce greenhouse gas emissions, there is a growing need to address power consumption in mobile networks. The proliferation of base stations and sophisticated services will significantly increase the power consumption of RANs. Reducing the power consumption of RAN operations is a critical challenge.

Fujitsu’s approach to RAN operational management

Fujitsu aims to tackle these challenges through “proactive automatic RAN optimization.” This approach involves real-time control of network equipment functions based on current and predicted network quality. By proactively addressing network quality degradation, Fujitsu ensures that users enjoy high-quality services while optimizing power consumption.

Proactive AI-driven network optimization offers several advantages. It automatically detects QoS degradation in real time and adjusts network resources accordingly, reducing excess resources and preventing QoS degradation. It also applies an evaluation index to QoE and predicts resource fluctuations and QoE scores, reducing power consumption while maintaining QoE.

RAN operational management architecture

Fujitsu’s architecture for achieving proactive automatic RAN optimization comprises four essential components:

• The QoS Monitoring and Prediction enables accurate forecasts of QoS variations
• The Real-Time QoE Monitoring allows for the detection of short-term QoE degradation that traditional QoS scores may not capture
• The Dynamic RAN Optimization controls O-DUs and O-CUs, ensuring they meet various performance requirements
• The Packet Analysis for QoE Monitoring collects real-time packet data and extracts relevant features for calculating current QoE scores

Fujitsu’s proactive automatic RAN optimization architecture aims to enhance end-user experiences by ensuring consistent network quality, addressing short-term degradation, optimizing performance, and responding swiftly to network issues. Users can expect improved reliability, reduced latency, and overall better satisfaction with the network services they rely on.

Advanced AI technologies by Fujitsu

It is imperative to leverage innovative AI technologies to ensure our networks are reliable and capable of making decisions that can be thoroughly understood and explained. The Fujitsu toolbox includes an array of advanced AI techniques, each designed to address specific challenges in RAN optimization. These technologies include Time Series Prediction, Wide Learning, and the innovative Constrained KPI-managing Multi-Agent Reinforcement Learning (CK-MARL). Together, they empower us to predict communication traffic patterns, estimate Quality of Experience (QoE) accurately, and orchestrate complex real-time optimizations on a large scale, paving the way for more efficient and intelligent mobile networks.

Fujitsu’s advanced RAN operational management solutions

Fujitsu’s advanced RAN operational management solutions address the evolving challenges in RAN operations. These solutions offer proactive and automatic optimization, ensuring high network quality while optimizing power consumption. By incorporating predictive AI technologies and real-time QoE monitoring, Fujitsu’s approach represents a significant step toward the intelligent and efficient management of Radio Access Networks.

Conclusion

Challenges in RAN operations, including network complexity due to service diversification and increased power consumption, necessitate proactive solutions. Fujitsu’s approach to proactive automatic RAN optimization, powered by advanced AI technologies, addresses these challenges by predicting and automatically optimizing network quality while optimizing power consumption. This approach ensures that users can enjoy high-quality services and paves the way for future efficient and intelligent management of RANs.