On the Combination of AI and Wireless Technologies: 3GPP Standardization Progress

Chen Sun,Tao Cui,Wenqi Zhang,Yingshuang Bai,Shuo Wang,Haojin Li
2024-06-17
Abstract:Combing Artificial Intelligence (AI) and wireless communication technologies has become one of the major technologies trends towards 2030. This includes using AI to improve the efficiency of the wireless transmission and supporting AI deployment with wireless networks. In this article, the latest progress of the Third Generation Partnership Project (3GPP) standards development is introduced. Concentrating on AI model distributed transfer and AI for Beam Management (BM) with wireless network, we introduce the latest studies and explain how the existing standards should be modified to incorporate the results from academia.
Networking and Internet Architecture,Artificial Intelligence
What problem does this paper attempt to address?
The paper primarily explores the integration of Artificial Intelligence (AI) with wireless communication technology and highlights the development progress of the Third Generation Partnership Project (3GPP) standards. Specifically, the paper attempts to address the following key issues: 1. **Distributed Transmission of AI Models**: - Utilizing AI to improve wireless transmission efficiency and how to support AI deployment through wireless networks. - Investigating the distributed transmission mechanisms of AI models across different devices, including model partitioning and transmission in Federated Learning (FL). 2. **Application of AI in Beam Management (BM)**: - Using AI technology to improve the beam management process, reduce system complexity, and enhance positioning performance in Non-Line-of-Sight (NLoS) scenarios. - Proposing AI-based beam prediction methods to reduce measurement overhead and latency, and improve beam switching efficiency. 3. **Development of 3GPP Standards**: - The paper provides a detailed introduction to the research progress on AI/ML model transmission in 3GPP standards, particularly the specification updates in Release 18 and 19. - Including how to support AI/ML operations through network automation functions (such as NWDAF) and how to implement AI-enhanced functions at the physical layer (such as CSI feedback, beam management, and positioning). 4. **Model Management and Performance Evaluation**: - Studying the specific details of management operations such as model activation, deactivation, updating, and selection. - Evaluating the performance of AI-enhanced functions, including the accuracy of compressed and predicted CSI, as well as the precision of beam prediction and positioning. In summary, this paper aims to address multiple challenges in wireless networks by combining AI technology with wireless communication technology, thereby enhancing the overall performance and efficiency of wireless communication systems.