Spectrograms of SDR Acquired Wireless Frames

Data Acquisition Scenario: Two smart phones are exchanging file transmissions using Bluetooth. While this is in progress, a Software Defined Radio (SDR) named ADALM PLUTO whose Rx port has JCG401 antenna, is programmed to capture 100 frames, each of length 30000 samples, at 2.4 GHz under the highest gain setting, with a baseband sample rate of 1 MHz. Each frame is cut into 5 sub-frames of 6000 samples each and a total of 100*5=500 sub-frames each of length 6000. Thus a complex data matrix of dimension 6000 rows and 500 columns is obtained is obtained. Secondly, similar to the above process, while WiFi based file transmission between a mobile phone and a laptop is in progress, the SDR is used to receive WiFi frames of 6000 rows and 500 columns. Thirdly, when both WiFi based file transfer and Bluetooth based file transfer are simultaneously happening between mobile-laptop and mobile-mobile respectively, the same SDR is used to capture frames of 6000 rows and 500 columns.

Dataset format: The 6000 length frames obtained in the above three data matrices are then used to obtain spectrograms dataset of dimension 224 by 224 by 1 by 1500, notated as variable X. The labels are arranged in a variable Y of dimension 1500 rows and 3 columns in a one-hot encoding scheme. The first 500 spectrograms correspond to WiFi, and the label is given as [1 0 0]. Spectrograms from 501 to 1000 correspond to Bluetooth and the label is given as [0 1 0]. Spectrograms 1001 to 1500 correspond to simultaneous WiFI and Bluetooth, and the label is given as [0 0 1].

Usage of the dataset: The dataset is a .mat file, with variables X and Y. Note that X has dimension 224 by 224 by 1 by 1500 and it comprises of 1500 spectrogram images each of dimension 224 by 224. Y has dimension 1500 by 3 and it consists of labels of the three classes of spectrograms namely
WiFi with class label [1 0 0]
Bluetooth with class label [0 1 0]
Simultaneous WiFi and Bluetooth with class label [0 0 1]

Applications: This dataset is an acquired real-world dataset from SDR ADALM PLUTO. It can be used for classification purposes to test different Machine Learning and Deep Learning based classification models. Automatic Waveform Classification Applications can use this dataset for testing the models that are trained with simulated datasets.