BLE RSS Measurements Dataset for Research on Accurate Indoor Positioning

Abstract

RSS-based indoor positioning is a consolidated research field for which several techniques have been proposed. Among them, Bluetooth Low Energy (BLE) beacons are a popular option for practical applications. This paper presents a new BLE RSS database that was created to aid in the development of new BLE RSS-based positioning methods and to encourage their reproducibility and comparability. The measurements were collected in two university zones: an area among bookshelves in a library and an area of an office space. Each zone had its own batch of deployed iBKS 105 beacons, configured to broadcast advertisements every 200 ms. The collection in the library zone was performed using three Android smartphones of different brands and models, with beacons broadcasting at −12 dBm transmission power, while in the other zone the collection was performed using of one those smartphones with beacons configured to advertise at the −4 dBm, −12 dBm and −20 dBm transmission powers. Supporting materials and scripts are provided along with the database, which annotate the BLE readings, provide details on the collection, the environment, and the BLE beacon deployments, ease the database usage, and introduce the reader to BLE RSS-based positioning and its challenges. The BLE RSS database and its supporting materials are available at the Zenodo repository under the open-source MIT license.

DL-IPS: Deep Learning Based Indoor Positioning System for Improved Accuracy

“…In addition, these Beacon markers are located in the topology of the corners of the rectangular test area for accessible real-life applications. The distance between the markers in the knit topology is positioned to be 5 meters or closer to minimize the estimated error rate [9]. As a result, a test area S was created in Fig.…”

A Meta-Review of Indoor Positioning Systems

“…Muñoz-Organero et al [104] tested the centroid of detected emitters, either giving or not predefined weights to each emitter. More recently, Mendoza-Silva et al [105] tested the centroid of the positions of the detected beacons in two environments, weighing each position according to the RSS of the beacon. The accuracy obtained using the weighted centroid was higher than using fingerprinting.…”

“…Fingerprinting has been an important approach for BLE positioning since its beginnings. As shown in Mendoza-Silva et al [105], fingerprinting is a better option than proximity or lateration when the number of BLE beacons is small. Zhang et al [109] used fingerprinting using two configurations of neural networks, Support Vector Machines (SVM) and k-Nearest Neighbors (kNN) with or without map enrichment with regression.…”

“…It is acknowledged that IPS based solely on WiFi fingerprinting can offer 2 m to 3 m of mean accuracy, although most common figures show 6 m to 7 m. BLE can provide better accuracies than WiFi, in the order of 1 m to 2 m, but larger values are also possible [105]. The variations in the previous numbers not only relate to specifics of the approach applied, but also to the characteristics of the environment and the collected training fingerprints.…”

Detecting and Correcting for Human Obstacles in BLE Trilateration Using Artificial Intelligence

“…Many proposed BLE RSSI-based ranging and fingerprinting systems often use RSSI from all three advertising channels together to obtain an aggregate signal [2][3][4]. However, the signals have been shown to exhibit different overall path-loss models and are affected by multipath fading and attenuation differently.…”