KT said on the 11th that it successfully completed a demonstration of an AI-based next-generation wireless transceiver technology in collaboration with Qualcomm Technologies and Rohde & Schwarz. The technology is an AI-based CSI (Channel State Information) compression method under discussion in 3GPP Release 20, in which AI learns the wireless environment between the device and the base station to improve the accuracy of channel state information.
KT and Qualcomm built a test environment by integrating a next-generation Snapdragon mobile platform with built-in AI capabilities and Rohde & Schwarz's AI base station system (CMX500 OBT). KT then applied an AI model that reflects its 5G network operation experience and base station configuration information to optimize it for a real commercial environment. As a result, channel state information accuracy improved and downlink performance increased by about 50%.
The demonstration showed that AI can substantially improve wireless performance itself and is significant in that it verified the effectiveness of 3GPP standards-based AI-RAN technology at the level of a commercial network.
KT plans to advance network operations AI technologies such as AI TEMS and ESKIMO while accelerating 6G standardization and the transition to AI-based autonomous operation networks through cooperation with global corporations.