Wireless World Model - 3D Point Clouds

Dataset Overview

This dataset provides three-dimensional (3D) point cloud representations of all environments used in the study “Wireless World Models for Future AI-Native Networks”.

The dataset contains reconstructed urban point clouds corresponding to the simulation and real-world measurement environments used in the project. These point clouds describe the geometric structure of the environments and can be used for visualization, environment modeling, and spatial analysis related to wireless propagation.

The dataset includes point clouds for the following urban environments:

Beijing Central Business District (CBD)

The Forbidden City, Beijing

Place de l’Étoile (Paris)

Wall Street, New York

Urban area of Munich

China Mobile International Information Port, Changping, Beijing

Each point cloud represents the 3D spatial structure of the corresponding environment and can be used to analyze the physical environment underlying the wireless channel datasets.

Data Generation Method

The point clouds were generated from geographic data sources used to reconstruct the urban environments for wireless channel simulation and field measurements. The reconstructed environments capture the large-scale geometry of buildings and urban structures, which are essential for realistic modeling of wireless signal propagation.

The point clouds represent sampled spatial points from the reconstructed 3D environment models.

File Organization

dataset/ ├── beijing_cbd.npy ├── forbidden_city_beijing.npy ├── paris_etoile.npy ├── wall_street_newyork.npy ├── munich_urban.npy └── changping_cmiiport.npy

Each file corresponds to the 3D point cloud of one environment.

Data Format

Each point cloud file is stored in NumPy (.npy) format.

The data array has the following shape:

(N, 3)

where

N : number of points in the point cloud3 : spatial coordinates (x, y, z)

Each row represents the 3D coordinates of a point in the reconstructed environment.

Usage

Download the point cloud files and place them into the following directory of the wireless_world_model_v1 code repository:

src/datasets/point_clouds

The point clouds can also be loaded directly using NumPy for visualization.