a reliable dataset,leveraging specific set of four optical cameras to provide ground truth with millimeters accuracy. The introduced dataset consists of RSSI values collected from five BLE sensors together with synchronized Inertial Measurement Unit (IMU) signals from the target’s mobile device.
Given the recent surge of interest on location-based services via BLE beacons, lack of a dataset with ground truth (actual labels) can be a significant obstacle for advancement of BLE-based indoor tracking/localiztion algorithms and research reproducibility. The paper takes a first step towards this goal and introduces the IoT-TD dataset, where the “Ground Truth Trajectories” are recorded in a synchronized fashion with the RSSI values together with IMU sensor measurements obtained, synchronously, from the moving target’s hand-held device. All three components of the dataset are time-stamped and pre-processed being available publicly for future BLE and PDR tracking algorithmic developments.
The experimental environment used to construct the IoT-TD dataset is a 3.5 meters to 3.5 meters area. Five BLE modules are used together with the built-in IMU sensor measurements of the user’s mobile device, recorded synchronously with RSSI values.
The following three different are implemented to gather RSSI values via BLE beacons, IMU measurements from the mobile device, and ground truth trajectories via the Vicon system:
- Rectangular Walking, where the user walks constantly on the sides of the rectangular area on a pre-defined path;
- Diagonal Walking, where the target walks along the diagonals of the area with constant velocity, and;
- Random Track, where the user walks randomly inside the surveillance region.