IMUWiFine: End-to-End Sequential Indoor Localization

Dataset Information:

The dataset used for training and evaluation is available separately at ISSAI and consists of train, test and validation folders. Specific file paths must be set within the train.py script. The following github repository provides the source code for the paper "End-to-End Sequential Indoor Localization Using Smartphone Inertial Sensors and WiFi" (Nurpeiissov et al., 2022). The code implements an end-to-end sequential indoor localization architecture using a PyTorch-based model. This architecture takes as input WiFi Received Signal Strength Indicators (RSSI) and Inertial Measurement Unit (IMU) readings from a smartphone and outputs estimated (x, y, z) coordinates. The model architecture comprises a stack of ReLU, LSTM, and regression layers. The dataset, available separately, is divided into training, validation, and testing sets. The code allows for training the model and automatic evaluation on the testing set. Repository: https://github.com/IS2AI/IMUWiFine

Citation:

@INPROCEEDINGS{9708854,
  author={Nurpeiissov, Mukhamet and Kuzdeuov, Askat and Assylkhanov, Aslan and Khassanov, Yerbolat and Varol, Huseyin Atakan},
  booktitle={2022 IEEE/SICE International Symposium on System Integration (SII)}, 
  title={End-to-End Sequential Indoor Localization Using Smartphone Inertial Sensors and WiFi}, 
  year={2022},
  pages={566-571},
  doi={10.1109/SII52469.2022.9708854}}