The AI-RAN Alliance is also working with tech partners to create four, distinct AI-RAN Alliance-endorsed labs
The AI-RAN Alliance announced today that it now has 75 members from 17 different countries. The Alliance is also launching two flagship initiatives and will showcase several AI-RAN technology demos during MWC Barcelona next week.
Announced during Mobile World Congress (MWC) Barcelona 2024, the AI-RAN Alliance is a group of technology and telecom leaders focused on the integration of artificial intelligence (AI) into cellular technology to further advance radio access network (RAN) technology and mobile networks. Alex Jinsung Choi, chair of the AI-RAN Alliance and principal fellow of SoftBank Corp.’s Research Institute of Advanced Technology, previously told RCR Wireless News that the group’s main mission is to “weave AI right into the fabric of the radio access network.”
This goal is being tackled from three angles — or working groups — within the Alliance: AI-for-RAN, which exploring how AI can enhance performance, capabilities and efficiency of radio access networks, such as improving spectral efficiency and energy efficiency; AI-and-RAN, which focuses on the integration of AI and RAN workloads on a common shared infrastructure to maximize utilization and enable new services; and AI-on-RAN, which looks at enabling the RAN to run AI and Gen-AI/LLM/VLMs/agentic AI applications by identifying the necessary network requirements, as well as benchmarks application performance on 5G and beyond.
Now for the new stuff: During a press briefing earlier this week, Choi was joined by NVIDIA AI-RAN and 6G Technology Evangelist Kuntal Chowdhury, who said that getting curated datasets for researchers and developers in the telecom field has been difficult. “And this is because of a number of things — it could be privacy, it could be the proprietary nature of that data, it could be that the data is not is really curated,” he continued, adding that another problem is the lack of a benchmarking framework for telco AI algorithms — an issue that the GSMA is also looking to address with the launch of the GSMA Open-Telco LLM Benchmarks.
For its part in solving this problem, the AI-RAN Alliance has introduced ‘Data-for-AI,’ an initiative it said aims to “define and implement a structured approach to data management and the data collection pipeline from sources such as real-time systems and simulators,” as well as ‘Test Methodology’ to develop standardized evaluation metrics for AI-driven RAN solutions.
The Alliance is also working with Keysight Technologies, Northeastern University, Singapore University of Technology and Design (SUTD) and VIAVI Solutions to create four, distinct AI-RAN Alliance-endorsed labs. “These labs will be used for data creation and benchmarking for all Alliance’s work. By doing so, we are not only addressing the AI and RAN aspects … but also creating the platform and foundation for innovation in a much more open and collaborative way,” said Chowdhury.
At next week’s MWC, the Alliance will conduct various demonstrations, including a showcase of how AI-driven RAN orchestration with Keysight and Northeastern University Open6G; a demo of how AI/ML-based higher-order modulations can deliver improved performance over square QAMs with NVIDIA, Samsung and VIAVI; and an examination of the potential of AI-enhanced RAN sensing for dynamic spectrum sharing with Northeastern University Open6G.
“We are all working together to shape the future of AI-powered RAN,” commented Choi the press briefing. He explained that some of the new Alliance additions include device chipmakers (like Qualcomm) and those working in the BSS/OSS domains. “And this is just the beginning; we have 50 more companies in the membership pipeline… This kind of scale is crucial because the AI-driven transformation in telecom is not something that one company, [or] even a handful of companies, can do alone. It takes an entire ecosystem working together — and we are building that ecosystem at speed,” he continued.
