Wi-Fi Crowdsourced Fingerprinting Dataset for Indoor Positioning

Abstract

Benchmark open-source Wi-Fi fingerprinting datasets for indoor positioning studies are still hard to find in the current literature and existing public repositories. This is unlike other research fields, such as the image processing field, where benchmark test images such as the Lenna image or Face Recognition Technology (FERET) databases exist, or the machine learning field, where huge datasets are available for example at the University of California Irvine (UCI) Machine Learning Repository. It is the purpose of this paper to present a new openly available Wi-Fi fingerprint dataset, comprised of 4648 fingerprints collected with 21 devices in a university building in Tampere, Finland, and to present some benchmark indoor positioning results using these data. The datasets and the benchmarking software are distributed under the open-source MIT license and can be found on the EU Zenodo repository.

Long-Term WiFi Fingerprinting Dataset for Research on Robust Indoor Positioning

“…A folder contains the datasets gathered for the collection month that the folder's name indicates. For each dataset, there are four files that store its RSS values, positions, times and identifiers sets, following a schema similar to that used in Lohan et al [10]. A file's name indicates the dataset kind ("trn" for training, "tst" for test), the dataset number, and which dataset's set represents ("rss", "crd", "tms" and "ids" for RSS values, positions, times and identifiers sets, respectively).…”

“…Reasons behind such contradiction include: (1) reporting the accuracy as mean positioning error distance; (2) testing the IPS in specific (probably controlled) environments; and (3) testing the IPS without considering temporal signals changes. The previous challenges have been addressed, e.g., by: (1) providing other metrics such as 75 percentile instead of the mean [8]; (2) providing databases [9,10]; and (3) periodically updating the IPS training data [11][12][13] or making the positioning method adaptable to signal changes [14,15]. Methods able to cope with temporal signal variation, such as those in Gu et al [14], Hayashi et al [15], are tested with measurements that allow the analysis of short-term signal variations occurred at known positions (e.g., seconds or minutes apart, caused by network devices dynamic behavior, network usage, and people movement) and also the analysis of long-term signal variations (e.g., days or months apart, caused by changes in network devices' configuration or environment' structure).…”

“…To foster reproducibility and comparability in indoor positioning research, several databases have recently been made available to the public [10,[16][17][18][19][20]. More specifically, Table 1 presents public databases that we have found available on-line and which can be used to train a WiFi RSS-based IPS.…”

Environmental Monitoring with Distributed Mesh Networks: An Overview and Practical Implementation Perspective for Urban Scenario

“…It consists of the following subgroups: (i) structural health [21]; (ii) light pollution monitoring [22]; (iii) waste management [23]; (iv) noise monitoring [24]; and (v) air pollution [25]. A massive section of this group is related to industrial control [26], aiming at: (i) indoor air quality monitoring [27], i.e., monitoring of toxic gas and oxygen levels inside chemical plants and office spaces to ensure safety; (ii) temperature monitoring [28], i.e., control of the temperature inside industrial and medical fridges with sensitive products; (iii) ozone level monitoring [29], i.e., monitoring of ozone levels inside food factories; and (iv) indoor positioning [30], i.e., indoor asset location utilizing active (ZigBee and Ultra-Wideband (UWB)) and passive (Radio Frequency Identification (RFID) and Near Field Communication (NFC)) tages. Nonetheless, security and emergency scenarios are also to be considered [31] as, for example: (i) perimeter access control [32], i.e., border surveillance and intrusion detection; (ii) dangerous liquid presence and leak detection [33,34], i.e., monitoring of the lower explosive limit of potentially dangerous gases and vapors; (iii) radiation level monitoring [35], i.e., real-time monitoring of radiation levels at nuclear facilities and surrounding areas; and (iv) explosive and hazardous gases in underground environments [36], i.e., continuous monitoring of the ambient characteristics of the mining environment.…”

Integration of positioning and activity context information for lifelog in urban city area

“…When the environment changes significantly (such as after a building renovation or moving of furniture), the database has to be rebuilt. 47 In addition, the disadvantages of the conventional method become more serious when Wi-Fi fingerprinting technology is deployed in a wide area. Our proposed method can determine the position of calibration points automatically when creating the database.…”