Segment-Wise Flow Matching for Vision-Aided mmWave V2I Beam Prediction

2025/11/25 by Can Zheng, Jiguang He, Zheng, Can +9
Engineering · Computer Science · #Millimeter-Wave Propagation and Modeling #Vehicular Ad Hoc Networks (VANETs) #Software-Defined Networks and 5G

paper · pdf · doi:10.48550/arxiv.2511.20265

Abstract

This paper proposes a vision-conditioned flow matching (FM) framework for beam prediction in millimeter-wave vehicle-to-infrastructure links. Instead of modeling discrete beam-index sequences, the proposed method learns the temporal evolution of normalized beam receive power vectors through a continuous vector field governed by an ordinary differential equation, enabling smooth dynamics and efficient sampling. By imposing FM over beam-state transitions and jointly optimizing beam prediction and flow consistency, the proposed framework provides a unified model for future beam prediction. Experimental results show that the proposed FM-based model significantly improves beam prediction performance over baselines, approaches the performance of large language model-based methods, and reduces predictor-side inference latency by about 6.9× on GPU and 2.8×10³× on CPU, respectively.

Citations

• Flow Matching-Based Autonomous Driving Planning with Advanced Interactive Behavior Modeling
• BeamLLM: Vision-Empowered mmWave Beam Prediction with Large Language Models
• Pyramidal Flow Matching for Efficient Video Generative Modeling
• Generative Pre-training for Speech with Flow Matching
• Multimodal Transformers for Wireless Communications: A Case Study in Beam Prediction
• Camera Based mmWave Beam Prediction: Towards Multi-Candidate Real-World Scenarios
• DeepSense 6G: A Large-Scale Real-World Multi-Modal Sensing and Communication Dataset
• Flow Matching for Generative Modeling
• Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow
• Computer Vision Aided Beam Tracking in A Real-World Millimeter Wave Deployment
• Millimeter Wave Base Stations with Cameras: Vision Aided Beam and\n Blockage Prediction
• Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry\n and Semantics
• Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks

Related

• BeamLLM: Vision-Empowered mmWave Beam Prediction with Large Language Models
• M2BeamLLM: Multimodal Sensing-empowered mmWave Beam Prediction with Large Language Models
• Integrated Sensing and Communications Towards Proactive Beamforming in mmWave V2I via Multi-Modal Feature Fusion (MMFF)
• LiDAR Aided Future Beam Prediction in Real-World Millimeter Wave V2I Communications
• Integrated Sensing and Communications for V2I Networks: Dynamic Predictive Beamforming for Extended Vehicle Targets