KT and Samsung Electronics said on November 5 that they used AI to optimize a wireless network and reduced connection drops on high-speed trains by about 29%. The companies compared performance before and after applying the technique on part of the Gyeongbu Line.
The trial covered a stretch between Godeok-myeon in Pyeongtaek and Maesong-myeon in Hwaseong. On trains running at 250 km/h or faster, devices hop between base stations at short intervals, and a late handover can disrupt data sessions and calls.
The two companies analyzed radio signal data and connection-failure logs with AI to identify quality drops that recurred across multiple sections. The main cause was devices holding on to the existing base station too long before switching to the next one. Settings were adjusted to trigger the handover earlier in the problem areas, which lowered the share of dropped connections during travel.