TY - GEN
T1 - A Lightweight Multimodal Beam Prediction Model with Hierarchical Attention for 6G V2X Network
AU - Lei, Jiahao
AU - Liu, Jiajia
AU - Wang, Jiadai
AU - Fu, Qiang
N1 - Publisher Copyright:
© 2026 IEEE.
PY - 2026
Y1 - 2026
N2 - The evolution of next-generation communication technologies necessitates accurate and efficient beam prediction to ensure high-quality links in Vehicle-to-Everything networks. Numerous studies have explored beam prediction schemes, which can be categorized into those based on blind beam training, channel state tracking, and side information. While existing schemes suffer from excessive parameters and high latency, this work introduces a lightweight multimodal model centered on a hierarchical attention mechanism. The model employs local attention to capture fine-grained vehicle dynamics and global attention to model long-term scene evolution. Extensive experiments demonstrate the model's superior robustness and accuracy across diverse traffic scenarios, significantly outperforming existing benchmarks.
AB - The evolution of next-generation communication technologies necessitates accurate and efficient beam prediction to ensure high-quality links in Vehicle-to-Everything networks. Numerous studies have explored beam prediction schemes, which can be categorized into those based on blind beam training, channel state tracking, and side information. While existing schemes suffer from excessive parameters and high latency, this work introduces a lightweight multimodal model centered on a hierarchical attention mechanism. The model employs local attention to capture fine-grained vehicle dynamics and global attention to model long-term scene evolution. Extensive experiments demonstrate the model's superior robustness and accuracy across diverse traffic scenarios, significantly outperforming existing benchmarks.
UR - https://www.scopus.com/pages/publications/105045387573
U2 - 10.1109/ICC59461.2026.11588024
DO - 10.1109/ICC59461.2026.11588024
M3 - 会议稿件
AN - SCOPUS:105045387573
T3 - IEEE International Conference on Communications
BT - ICC 2026 - IEEE International Conference on Communications, Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2026 IEEE International Conference on Communications, ICC 2026
Y2 - 24 May 2026 through 28 May 2026
ER -