High precision indoor localization is a key enabler for various tasks in health care, industrial production or networking. While classical time-of-arrival (ToA) approaches fail in none-line-of-sight (NLoS) situations, fingerprinting based methods can achieve high localization accuracies under harsh radio propagation conditions.
Data acquisition
We recorded two different data sets: a training data set and a test data set, where one burst includes the six synchronized channel impulse responses (CIRs) and time-of-flight (ToF) measurements. Both data sets were recorded independently on different trajectories within the same area. We designed a complex indoor environment with walls, that reflect radio signals on the inner side (iron surface) and absorb them at the outside (black surface). The transceivers, indicated as green dots, are placed at the edges of the recording area. The walls, indicated in red, are placed to block the line-of-sight (LoS) between the anchors and the robot platform, which causes ranging errors of the UWB radio system leading to high localization errors using classical positioning approaches.
Sensors
Hardware
For data recording we used the Wismit platform, which can be used for the evaluation of different positioning technologies. For ultra-wideband (UWB) localization the Decwave DW1000 is utilized. We configured the system for ToF acquisition as a two-way ranging setup with a bandwidth of 499.2 MHz at a center frequency of 4 GHz with a recording frequency of 3 Hz.
Sensor data
The sensors raw data is provided including ToF measurements and the corresponding CIRs. The data format is HDF5 with the following columns:
A_ID: Unique identifier for the stationary transceiver module
B_ID: Unique identifier of the burst
TD: Estimated ToF of the UWB module
TD_OFFSET: Relative ToA within the window of the CIR
CIR_R: Real part of the CIR
CIR_I: Imaginary part of the CIR
POS_X: X-coordinate of the ground truth position
POS_Y: Y- coordinate of the ground truth position
Data acquisition
We recorded two different datasets a training data set and a test data set on different trajectories within the same environment. The measurements include time-difference-of-arrival (TDoA) and corresponding CIRs. We created a typical industrial setup with reflective/absorbing walls (red), an industrial truck (orange), a forklift (gray), and small (blue) and large shelves (purple) to block the LoS to the receiver and to create dense multipath propagation. The transceivers (green) are placed at the edges of the recording area at a height of 6-7 m. The receiver is placed at a height of 1.95 m on a handcart, which is moved by a person.
Sensors
Hardware
As radio setup we used a 5G downlink TDoA radio system with six commercial off-the-shelf software-defined-radio BS. The radio system has a bandwidth of 100 MHz and a center frequency of 3.7 GHz. The BS transmit power is set to 20 dBm. All BSs are highly synchronized by means of a common signal generator. The recording frequency is 6.6 Hz.
Sensor data
The sensors raw data is provided including TDoA measurements and the corresponding CIRs. The data format is HDF5 with the following columns:
A_ID: Unique identifier for the stationary transceiver module
B_ID: Unique identifier of the burst
TD: Estimated TDoA using the shortest ToA as reference
TD_OFFSET: Relative ToA within the window of the CIR
CIR_R: Real part of the CIR
CIR_I: Imaginary part of the CIR
POS_X: X-coordinate of the ground truth position
POS_Y: Y- coordinate of the ground truth position
Fraunhofer Institute for Integrated Circuits IIS