Citing artificial intelligence (AI) and 5G as cornerstone technologies of this decade, a new whitepaper released today by InterDigital, Inc. and written by ABI Research details the integral importance of AI to achieve 5G’s potential. The whitepaper, AI is the Key for 5G Success, examines the network domains and standards efforts where AI and ML are being integrated to automate critical 5G operations.

According to the whitepaper, 5G promises to support multiple spectral assets, from low-band to millimeter Wave frequencies, and a proliferation of diverse IoT, consumer, and enterprise use cases, therefore requiring complex AI algorithms to achieve desired network function and quality user experience. Demand for advanced AI algorithms to enhance network planning and maintenance also is expected to grow as industry considers energy costs and network sustainability and prepares for future use cases and services.

“AI/ML will deliver benefits to nearly every layer of the 5G network domain, from the RAN and core networks to business and operational applications over top. While AI/ML is increasingly integrated within the business and operational layers of 5G networks today, tech evolution will see greater AI/ML impact within network foundations like the RAN,” said  VP and Head of Wireless Labs at InterDigital, Milind Kulkarni. “This new frontier of AI/ ML integration will require new R&D in machine learning algorithms, and InterDigital is proud to drive research in this rich area of innovation and standardization.”

The whitepaper describes key use cases for AI in 5G networks, including the following:

Network optimization through Dynamic Traffic Steering uses intelligent AI/ML solutions to improve user experience as the network manages traffic across different access technologies, including 4G, Wi-Fi and 5G.

The RAN is beginning to leverage AI algorithms to optimize Massive MIMO to calculate the optimal angle, position, beamwidth, and other parameters that help drive the massive capacity increase from 4G to 5G networks.

RAN Intelligence Controller (RIC) and Service Management and Orchestration (SMO) are promising network features introduced in Open RAN networks that can drive significant value by eliminating market barriers to open interfaces from third parties and allowing new entrants to apply their innovative algorithms to telecoms networks.

AI models and algorithms are being tapped to reduce network costs in a variety of domains, including bridging varying network vendor equipment and management models, leading real-time resource management, predicting future network requirements to optimize procurement decisions, and aiding with network planning and deployment.

AI/ML capabilities are critical for cellular service providers to embrace full closed-loop automation and progress towards applications that can make and implement self-driven decisions for the network, with minimal human input. This approach requires a variety of KPIs from across the mobile ecosystem and an end-to-end orchestrated and tightly integrated network.

“AI is not a new concept in cellular and telecom networks – in fact it has been used to optimize many parts of the network stack and business processes for years. However, network automation using AI/ML is now becoming vital to manage the number of devices, increasing amounts of traffic and a plethora of use cases in the network domain,” said Senior Research Director at ABI Research, Dimitris Mavrakis. “The proliferation of AI/ML in cellular networks is now creating the foundation upon which newer generations will be built, including 5G Advanced and 6G, which will introduce completely new types of service innovation and consumer experiences.”

To achieve greater network automation, cellular service providers increasingly use standards-based architectures and open APIs within AI/ML-based automation platforms to overcome extensive customization and siloed processes in the network stack. 3GPP is exploring the use of AI in the 5G network in Release 18, and InterDigital is actively supporting the effort to investigate the benefits of augmenting the New Radio Air Interface with AI/ML models to enhance performance and reduce complexity. Additionally, the ORAN Alliance has established reference architecture for RIC and SMO network elements to leverage AI algorithms in order to maintain and optimize the network, while ETSI has developed the Experiential Network (ENI) framework for AI in telecoms networks to create closed-loop management systems to enhance network operations.

As the proliferation of devices, spectrum bands, data traffic, and infrastructure vendors makes 5G networks increasingly complex, AI is an important tool to automate and manage the critical network operations today, and into the future.