Scene2Wave is a multimodal wireless channel dataset designed to provide high-precision, time-synchronized multimodal data for wireless channel representation learning, path-level channel impulse response and channel state information estimation and prediction, perceived channel feature extraction, and multimodal synchronization fusion research. Version 1.1.0 of this dataset contains 300 officially validated samples, covering multiple CARLA towns, different motion states, and radio configurations. It provides multi-view synchronized sensor and wireless channel data for autonomous intelligent vehicles (CAVs) and roadside units (RSUs). The main modalities include multi-view cameras, bird's-eye view, LiDAR, radar, inertial measurement unit (IMU), Global Navigation Satellite System (GNSS), vehicle/roadside pose, path-level channel impulse response (CIR), and channel state information (CSI). Dataset composition:

• Scene distribution: Covering four towns: CARLA Town01, Town02, Town03, and Town15, with 25 samples from each town.
• Motion states: Includes five vehicle speeds: static, 10 km/h, 20 km/h, 40 km/h, and 60 km/h, with 20 samples for each state.
• Propagation Model: Includes four wireless propagation configurations: core 40, multipath 20, scattering_mild 20, and scattering_medium 20.