Wireless World Model - Field Dataset

Dataset Overview This dataset provides real-world wireless channel measurements for the study “Wireless World Models for Future AI-Native Networks”. It complements the simulation datasets by providing real-world uplink channel state information (CSI) collected from outdoor field trials. The dataset enables the evaluation of learning-based wireless channel modeling and prediction methods under practical deployment conditions. The dataset consists of CSI measurements collected along several user equipment (UE) trajectories. The measurements are synchronized with UE spatial coordinates to form time-continuous CSI–position pairs. Two subsets are provided: field_train: training data containing two UE trajectories. field_test: testing data containing an independent UE trajectory for evaluation. Each trajectory (trace) contains approximately 20,000 CSI samples. Data Generation Method The dataset was collected from real-world field trials using a 6G prototype system developed by the China Mobile Research Institute. The prototype integrates advanced 6G transmission technologies and provides a high-fidelity environment for iterative technology validation and wireless data acquisition. The dataset consists of uplink CSI measurements based on Sounding Reference Signal (SRS) transmissions. Outdoor measurements were conducted at the China Mobile International Information Port in Beijing, China. The measurement system operated at a 6.6 GHz carrier frequency with a 400 MHz bandwidth. During data collection, the UE moved along predefined trajectories while continuously transmitting SRS signals. The received CSI was recorded together with the corresponding UE positions to construct synchronized CSI–trajectory data. File Organization dataset/ ├── train_field.zip │ ├── H/ │ │ ├── samples_H_changping_trace0_5kmh.npy │ │ └── samples_H_changping_trace1_5kmh.npy │ └── pos/ │ ├── samples_pos_changping_trace0_5kmh.npy │ └── samples_pos_changping_trace1_5kmh.npy │ └── test_field.zip ├── H/ │ └── samples_H_changping_trace2_5kmh.npy └── pos/ └── samples_pos_changping_trace2_5kmh.npy train_field.zip contains two measurement trajectories (trace0 and trace1) used for model training. test_field.zip contains an independent trajectory (trace2) used for evaluation. Each trajectory contains approximately 20,000 CSI samples. Data Format Each dataset file is stored in NumPy (.npy) format. The CSI data files samples_H_*.npy contain the measured uplink CSI matrices. The position data file samples_pos_*.npy stores the UE trajectory information associated with each CSI sample. Each position file contains dictionary entries corresponding to CSI samples, including the UE trajectory and measurement metadata. The CSI data file samples_H_*.npy has the following shape: (samples, 2, 16, 32, 32) where samples : number of channel samples2 : real and imaginary components16 : temporal snapshots32 : base station antennas features32 : user equipment antennas and frequency subbands features The real and imaginary parts are stored separately along the second dimension. The position data file samples_pos_xxx_bsx_xxkmh_2048_seedxxxxxxxxxx.npy has the shape: (samples,) The file contains approximately 20,000 dictionary entries, where each dictionary corresponds to one CSI sample. Each dictionary includes the following keys: ue_traj: the trajectory of the user equipment (UE), storing the UE positions at 16 time steps. cityname: the name of the urban scenario in which the data are generated. bs_pos: the spatial coordinates of the base station used in the simulation. Usage Unzip the dataset files into the directory src/datasets/raw_data/ of the wireless_world_model_v1 repository as follows: src/datasets/raw_data/train_fieldsrc/datasets/raw_data/test_field To preprocess the datasets, run preprocess_example_script.py and set the parameter: RAW_DATASET_NAME='train_field' or RAW_DATASET_NAME='test_field' For further details on dataset preprocessing and usage, please refer to the repository README.md.