The UPC-LPWAN-1 dataset is a comprehensive collection of Sub-GHz Low-Power Wide-Area Network (LPWAN) signal recordings, meticulously curated to address the need for standardized datasets in the evaluation and comparison of machine learning models for system/signal classification. This dataset contains a diverse array of transmission schemes including both single Carrier and multi Carrier transmision modes of 802.15.4g-2020 (WiSUN), IEEE 802.15.4g-2012, IEEE 802.11ah (WiFi HaLow), LoRa and SigFox resulting in a total of 40 classes. For this purpose, multiple commercial boards and devices have been used, generating a wide range of signals that cover all possible configurations. UPC-LPWAN-1 is publicly available to the scientific community, offering both raw recordings that can be freely modified for various research purposes and a post-processed version ready to use and with a variety of signal-to-noise ratios. This dataset aims at playing a significant contribution to the field of LPWAN signal analysis.

Raw recordings are in HDF5 format in .mat files. This format is compatible with popular programming languages such as Python, R, C, Matlab, etc. Processed data has been serialized in .pkl format, so it can be read with Python