Research Article
VPAO-SIM: Vision-Based Perception and Alternating Optimization for SIM Beamforming in Vehicular Networks
@INPROCEEDINGS{10.1007/978-3-032-33662-0_8, author={Shuyi Chen and Jiaye Wang and Lin Ma and Liang Ye and Yu’e Jiang and Yao Xu and Weixiao Meng}, title={VPAO-SIM: Vision-Based Perception and Alternating Optimization for SIM Beamforming in Vehicular Networks}, proceedings={Communications and Networking. 20th EAI International Conference, ChinaCom 2025, Xi’an, China, November 8--9, 2025, Proceedings}, proceedings_a={CHINACOM}, year={2026}, month={10}, keywords={Stacked intelligent metasurface (SIM) vision-based beamforming alternating optimization vehicular networks interference suppression}, doi={10.1007/978-3-032-33662-0_8} }- Shuyi Chen
Jiaye Wang
Lin Ma
Liang Ye
Yu’e Jiang
Yao Xu
Weixiao Meng
Year: 2026
VPAO-SIM: Vision-Based Perception and Alternating Optimization for SIM Beamforming in Vehicular Networks
CHINACOM
Springer
DOI: 10.1007/978-3-032-33662-0_8
Abstract
Next-generation vehicular networks require high-throughput, low-latency, and adaptive beamforming to support intelligent transportation systems. Stacked Intelligent Metasurfaces (SIM), as a transmissive alternative to Reconfigurable Intelligent Surfaces (RIS), offer passive beamforming with enhanced spatial control. However, challenges such as the difficulty of acquiring accurate Channel State Information (CSI) in dynamic environments and the limitations of low-resolution phase control hinder their practical deployment. To address these issues, we propose VPAO-SIM—a vision-based perception and alternating optimization framework for SIM beamforming. VPAO-SIM employs monocular 3D vision to estimate real-time angles of arrival, which guide a quantization-aware alternating optimization algorithm to configure SIM layers for hardware-constrained Linearly Constrained Minimum Variance (LCMV) beamforming. Simulation results show that VPAO-SIM achieves near-optimal target gain and interference suppression with few-bit quantization. The method is robust to vision errors and scales to multi-user scenarios, offering a practical solution for perception-driven beamforming in dynamic vehicular networks.