A multisource and multivariate dataset for indoor

A multisource and multivariate dataset for indoor localization methods based on WLAN and geomagnetic field fingerprinting Antonino Crivello*

P. Barsocchi, A. Crivello, D. La Rosa, F. Palumbo ISTI – CNR, Pisa - Italy

1.  Why spending effort ? 2.  Data Acquisition 3.  Data Set Description 4.  Discussion and conclusion

Wi-Fi probes as digital crumbs for crowd localisation – IPIN 2016, Spain | A. Crivello - ISTI, CNR – DIISM, Univ. Of Siena

1.1 1.

Why spending effort ?

•  Many papers in the literature trying to solve the indoor localisation issue •  Each approach presents algorithms and results using its own dataset based on Wi-Fi or Geo-Magnetic field measurement Wi-Fi probes as digital crumbs for crowd localisation – IPIN 2016, Spain | A. Crivello - ISTI, CNR – DIISM, Univ. Of Siena

1.2 1.

State-of-the-art

•  The lack of a common dataset and frameworks to compare and evaluate solutions is the main drawback in this field


1.3 1.

Novelty

•  J. Torres-Sospedra et al. presented two different public dataset (Wi-Fi [1] and GeoMag [2]) •  The dataset presented addresses these issues by gathering, in the same environment, Wi-Fi fingerprints, geomagnetic field, and inertial sensors •  The dataset is temporary available at http://wnet.isti.cnr.it/software/Ipin2016Dataset.zip (as soon as possible the dataset will be moved to the official UCI repository) [1] J. Torres-Sospedra, R. Montoliu, A. Martinez-Uso, J. P. Avariento, T. J. Arnau, M. Benedito-Bordonau, and J. Huerta, “Ujiindoorloc: A new multi-buildingandmulti-floordatabaseforwlanfingerprint-basedindoor localization problems,” in Indoor Positioning and Indoor Navigation (IPIN), 2014 [2] J.Torres-Sospedra,D.Rambla,R.Montoliu,O.Belmonte,andJ.Huerta, “Ujiindoorlocmag:Anewdatabaseformagneticfield-basedlocalization problems,” in Indoor Positioning and Indoor Navigation (IPIN), 2015 Wi-Fi probes as digital crumbs for crowd localisation – IPIN 2016, Spain | A. Crivello - ISTI, CNR – DIISM, Univ. Of Siena

1.  Why spend effort ? 2.  Data Acquisition 3.  Data Set Description 4.  Discussion and conclusion


2.1

Data Acquisition •  Two campaigns performed •  Both datasets collected cover a surface of 185.12 m2 •  Wearing two devices simultaneously: smartphone and smartwatch


2.2

Data Acquisition

•  The place in which the data collection was carried out is an indoor office environment composed of two rooms, two corridors and one small entrance hall. •  The points were equally spaced by 60 cm in both directions in order to uniformly cover the interested area. Wi-Fi probes as digital crumbs for crowd localisation – IPIN 2016, Spain | A. Crivello - ISTI, CNR – DIISM, Univ. Of Siena

2.3

Data Set Acquisition Parameters collected for each device


1.  Why spend effort ? 2.  Data Acquisition 3.  Data Set description 4.  Discussion and conclusion


3.1

Data Set description

•  Force applied to a device (x,y,z) •  Ambient geo-magnetic field (x,y,z) •  Degrees of rotation that a device makes around (x,y,z) •  Device’s rate of rotation around x,y,z •  Wi-Fi access points (APs): •  WiFi network name •  AP MAC address •  RSSI


3.2


•  The reference time used by the smartwatch is synchronized with the smartphone before starting each data acquisition session •  Sensors output is sampled with a frequency of 10 Hz


3.3

Data Set description •  The records of both datasets have been captured on 325 different points (dots on the map)


3.4


• 

WLAN: 650 records, 325 records retrieved during the first campaign and 325 during the second one

• 

Geomagnetic dataset - two different tables •  Smartphone sensors table •  Smartwatch sensors table


1.  Why spend effort ? 2.  Data Acquisition 3.  Data Set description 4.  Discussion and conclusion


4.1

Discussion - WLAN •  Place ID and SSID detected


4.2

Discussion - GeoMag

•  The cross-correlation function between the two magnitude signals. This confirms that the magnetic field measured remains almost constant over time


4.3

Discussion - GeoMag

•  Heat maps showing the data collected in different parts of the covered environment from


4.4

Conclusion

•  The space granularity offered by the geo-magnetic field surv is less than the one offered by the RSSs from Wi-Fi access points, that can be used to enhance the overall localization accuracy in hybrid methods

•  The presence of Wi-Fi RSSs and geo-magnetic field values, together with the map of the monitored environment, opens various possibilities for fingerprinting-based ILSs, helping researchers in the off-line collection phase.


4.5

Summary

•  lack of a common dataset •  Our dataset, in the same environment, collects Wi-Fi fingerprints, geomagnetic field, and inertial sensors •  Data gathered by smartphone and smartwatch •  Future works: compare how different techniques can benefit from fusing information

THANKS

[email protected]
