Federated Learning: mmWave MIMO radar dataset for testing
The database contains the raw range-azimuth measurements obtained from mmWave MIMO radars (IWR1843BOOST http://www.ti.com/tool/IWR1843BOOST) deployed in different positions around a robotic manipulator. Data has been collected inside a workspace environment. The data is used to train a machine learning model for the detection of a human operator placed in different positions (see the image). The Python code (use the link to download) uses the above mentioned data to implement decentralized federated learning stages via consensus and optimize the training loss and latency.
The database that contains the raw range-azimuth measurements obtained from mmWave MIMO radars inside a Human-Robot (HR) workspace environment.
The database contains 5 data structures:
i) mmwave_data_test has dimension 900 x 256 x 63. Contains 900 FFT range-azimuth measurements of size 256 x 63: 256-point range samples corresponding to a max range of 11m (min range of 0.5m) and 63 angle bins, corresponding to DOA ranging from -75 to +75 degree. These data are used for testing (validation database). The corresponding labels are in label_test. Each label (from 0 to 5) corresponds to one of the 6 positions (from 1 to 6) of the operator as detailed in the image attached.
ii) mmwave_data_train has dimension 900 x 256 x 63. Contains 900 FFT range-azimuth measurements used for training. The corresponding labels are in label_train.
iii) label_test with dimension 900 x 1, contains the true labels for test data (mmwave_data_test), namely classes (true labels) correspond to integers from 0 to 5.
iv) label_train with dimension 900 x 1, contains the true labels for train data (mmwave_data_train), namely classes (true labels) correspond to integers from 0 to 5.
v) p (1 x 900) contains the chosen random permutation for data partition among nodes/device and federated learnig simulation (see python code).