Data set was created using SNPN-UWB board with DecaWave DWM1000 UWB radio module.
Measurements were taken on 7 different indoor locations:
- Office1
- Office2
- Small appartment
- Small workshop
- Kitchen with a living room
- Bedroom
- Boiler room.
In every indoor location 3000 LOS samples and 3000 NLOS samples were taken. Different locations were choosen to prevent building of location-specific LOS and NLOS models. All together 42000 samples were taken: 21000 for LOS and 21000 for NLOS channel condition. To make data set ready for building LOS and NLOS models, samples are randomized to prevent overfitting of a model to particular places. For measurements two UWB nodes were used: one node as an anchor and the second node as a tag. Only traces of LOS and NLOS channel measurements were taken without any reference positioning (this data set is not appropriate for localization evaluation).
Folder with data set is organized as follows:
+ code
|____ uwb_dataset.py
+ dataset
|____ uwb_dataset_part1.csv
|____ uwb_dataset_part2.csv
|____ uwb_dataset_part3.csv
|____ uwb_dataset_part4.csv
|____ uwb_dataset_part5.csv
|____ uwb_dataset_part6.csv
|____ uwb_dataset_part7.csv
Whole data set is randomized and later split into 7 smaller files.
First line in every data set file is a header with column names. Elements of every sample are:
- NLOS (1 if NLOS, 0 if LOS)
- Measured range (time of flight)
- FP_IDX (index of detected first path element in channel impulse response (CIR) accumulator: in data set it can be accessed by first_path_index+15)
- FP_AMP1 (first path amplitude - part1) look in user manual
- FP_AMP2 (first path amplitude - part2) look in user manual
- FP_AMP3 (first path amplitude - part3) look in user manual
- STDEV_NOISE (standard deviation of noise)
- CIR_PWR (total channel impulse response power)
- MAX_NOISE (maximum value of noise)
- RXPACC (received RX preamble symbols)
- CH (channel number)
- FRAME_LEN (length of frame)
- PREAM_LEN (preamble length)
- BITRATE
- PRFR (pulse repetition frequency rate in MHz)
- CIR (absolute value of channel impulse response: 1016 samples with 1 nanosecond resolution)
To import data set data into Python environment, uwb_dataset.py script from folder code can be used. The CIR data still needs to be divided by number of acquired RX preamble samples (RX_PACC).
import uwb_dataset
# import raw data
data = uwb_dataset.import_from_files()
# divide CIR by RX preable count (get CIR of single preamble pulse)
# item[2] represents number of acquired preamble symbols
for item in data:
item[15:] = item[15:]/float(item[2])
print(data)
If you are using our data set in your research, citation of the following paper would be greatly appreciated.
Author of UWB LOS and NLOS Data Set and corresponding Python scripts is Klemen Bregar, [email protected].
Copyright (C) 2017 SensorLab, Jožef Stefan Institute http://sensorlab.ijs.si
The research leading to these results has received funding from the European Horizon 2020 Programme project eWINE under grant agreement No. 688116.