Background
After RAN#105, the work item on Artificial Intelligence (AI)/Machine Learning (ML) for NR Air Interface (NR_AIML_air; refer to RP-242399 for details) was revised to target enhancing 5G network performance. This evolution builds on prior studies to specify the signaling and protocols needed for AI/ML applications like beam management, positioning, and CSI feedback, ensuring a seamless operation between the network and user equipment (UE).
In the previous RAN2#127 meeting, delegates made several key agreements under agenda item 8.1.2.2 – LCM (Life Cycle Management) for the UE-sided model for Beam Management. These included:
- Definitions: Supported functionalities are those reported by the UE, applicable functionalities are ready for use, and activated functionalities perform inference.
- Configuration Process: The network queries and configures the UE’s functionalities based on conditions, with the UE reporting applicable functionalities when requested or when they change.
- Activation: The network provides inference configuration, and functionalities are activated, with flexibility for multiple activations based on the network setup.
These agreements form the foundation for the upcoming RAN2#127-bis meeting, where further technical challenges will be addressed to define standards for efficient coordination between the UE and network for AI/ML functionalities in beam management.
Overview of Proposals for RAN2#127-bis
Before attending the RAN2#127-bis meeting, RAN2 delegates must review the proposals submitted in TDocs to 3GPP. Specifically, in agenda item 8.1.1.2 – LCM for UE-sided model for Beam Management use case, 201 proposals across 29 TDocs contributed by 29 sources will shape future standards for AI/ML functionalities in NR systems.
Hot Topics
- 65 proposals (32%) on UE Training NW : focus on improving UE model training processes for efficient network interactions. Enhanced UE training is crucial for maintaining optimal performance and adapting to network changes.
- 59 proposals (29%) on Reporting Functionality : aim to improve the efficiency and accuracy of reporting mechanisms for applicable functionalities, which are vital for maintaining high-quality network performance and data analysis.
- 33 proposals (16%) on AI ML Functionality : target integrating AI/ML into network functionalities, enhancing automation, decision-making, and overall network intelligence.
- 44 proposals (21.9%) on Functionalities Agree Configured and Monitoring Performance RAN1.
Notable Signals
- 29 sources contributed proposals, with Qualcomm and Ericsson each contributing over 13 proposals, while the remaining 27 sources averaged 6.4 proposals each.
- Qualcomm’s 14 proposals in R2-2408390 focused primarily on UE Training NW and AI ML Functionality.
- Ericsson’s 13 proposals in R2-2409103 contributed significantly to UE Training NW and Monitoring Performance RAN1.
Understanding these insights allows RAN2 delegates to prioritize key discussion areas, anticipate meeting directions, and prepare effectively.
The Workload Challenge
The sheer volume of proposals—201 in total—presents a significant challenge for RAN2 delegates. Evaluating each proposal is crucial but also demanding, given the limited time before the meeting.
These challenges underscore the need for practical tools and insights to streamline the review process and help delegates set clear, impactful standards.
Introducing Proposal Zone
To help delegates navigate these complexities, we introduce Proposal Zone—an intuitive tool designed to simplify proposal analysis and enhance decision-making efficiency. Proposal Zone provides GPT-powered analysis, offering tags, topic summaries, and sentiment insights for each proposal. With Proposal Zone, delegates can:
- Quickly access relevant proposals by agenda item.
- Gain a comprehensive overview of key topics, trends, and sentiment.
- Effectively evaluate proposals, identifying potential allies or competitors.
Using Proposal Zone, delegates can streamline their preparation, ensuring they have the information they need to engage in productive discussions and set meaningful standards.
Explore how Proposal Zone can help you manage your workload effectively and contribute confidently to RAN2#127-bis. Visit https://ixi.wispro.com for more information.
Conclusion
As we approach this critical meeting, leveraging insights and tools can significantly impact how RAN2 delegates shape the future of AI/ML for NR Air Interface.