Samsung R&D Institute China-Beijing (SRC-B)’s Advanced Research & Standard Team won first place in the IMT-2030 (6G) AI Competition. This competition was organized by the IMT-2030 (6G) Promotion Group and sponsored by OPPO. Guided by the Ministry of Industry and Information Technology (MIIT) of China, the IMT-2030 (6G) Promotion Group serves as a key platform for advancing China's 6G research and development, as well as international cooperation, representing the forefront of the country’s 6G technology development. The competition aims to promote 6G and B6G technologies, explore the role of AI in addressing key technical challenges, and foster technological breakthroughs to secure leadership in the industry.

The topic of this IMT-2030 6G competition is “6G/B6G Native AI: ‘Semantic Control’ System Design.” This competition topic was proposed against the backdrop of current limitations arising from the modular architecture of existing wireless communication systems, where individual components undergo independent optimization. This approach inherently constrains achievable physical layer performance. Consequently, a systematic approach is urgently needed to innovatively redesign both data transmission links and their control mechanisms using AI. Participants are encouraged to design high-performance AI-based data transmission links and control mechanisms to enhance the overall system transmission gain, thereby exploring how AI technology can empower wireless communication systems from a systemic perspective.

SRC-B proposed a solution featuring adaptive data transmission links and control mechanisms based on channel characteristics. Its core technologies encompass ensemble learning, hybrid signal-to-noise ratio (SNR) enhancement, adaptive repetitive transmission, and differentiable accuracy loss. Under diverse channel conditions (Uplink SNR: -20 to 0 dB, Downlink SNR: -20 to 20 dB), this solution significantly outperformed baseline algorithms. It ultimately achieved first place with a clear advantage, demonstrated by reducing the bit error rate (BER) by 1 percentage point compared to the second-place solution.

SRC-B has consistently prioritized exploring the integration of AI and wireless communications, developing numerous native AI-powered solutions specifically designed for wireless systems. These innovations have garnered repeated international accolades. SRC-B’s expertise spans the entire communication stack—from RF and physical layer design to network optimization and source coding—while maintaining a dedicated commitment to advancing the wireless-AI convergence field through sustained innovation and groundbreaking contributions.