13:30–14:20 · 第一場
黃楚翔
國立臺灣大學 助理教授
曾任 Qualcomm Technologies 3GPP RAN 工作小組代表,參與標準會議及行動通訊產品研發;研究領域包括新世代無線通訊系統、標準制定與人工智慧/機器學習。
講題:AI in 6G Communications: New Use Cases for Interface Standardization
演講摘要
The evolution toward 6G communications necessitates an intelligent air interface capable of addressing complex, non-linear wireless challenges that traditional mathematical models struggle to resolve. This presentation outlines the 3GPP 6G study items focused on AI/ML-enhanced Radio Access Networks (RAN), specifically emphasizing use cases for interface standardization. Central to this development are several critical use cases identified for 6G interface design covered in this presentation. We first introduce the use cases in channel state information (CSI) and estimation, which focusing on the reference signal and reporting mechanism design to utilize neural networks for non-linear channel estimation and CSI encoding, decoding, and compression to reduce feedback overhead while improving reconstruction accuracy. Along this direction, we can first investigate the application of AI on Demodulation Reference Signal (DMRS) enhancement, cross-frequency CSI prediction. Power amplifier (PA) non-linearity compensation is another interesting use case discussed in 6G to leverage the strong capability of AI to learn and compensate the highly non-linear distortion in near saturation region of PA to further explore the high power transmission region of uplink.