Dataset for "Self-Supervised Learning for Spectrogram-Based Identification of Mobile Communications Technologies"

These spectogram datasets was created using the following datasets: POWDER - https://genesys-lab.org/powder, ICARUS - https://genesys-lab.org/ICARUS, Panoradio - https://panoradio-sdr.de/radio-signal-classification-dataset/, Real-world Radar and LTE Signals datasets - https://genesys-lab.org/CBRS, MATLAB - https://www.mathworks.com/help/comm/ug/spectrum-sensing-with-deep-learning-to-identify-5g-and-lte-signals.html. The "spectogram_rgb" folder contains the dataset used for pretraining the BYOL model. It contains all the datasets used. The "dataset_no_powder_rgb" folder contains the reduced dataset used for pretraining the BYOL model. The POWDER dataset is excluded. The "dataset_no_panoradio_rgb" folder contains the reduced dataset used for pretraining the BYOL model. The Panoradio dataset is excluded. The "dataset_spectogram" folder contais the MATLAB dataset used for finetuning. It is split into train, validation and test. Contains the spectogram image and its respective mask. It simulates LTE and NR signals. The "fragmented_data" and "fragmented_data_2" folder contains the MATLAB train dataset but fragmented. The spectograms are size: 256x256.