A comparative analysis of N-Nearest Neighbors (N3) and Binned Nearest Neighbors (BNN) algorithms for indoor localization

Serpil Ustebay (1) , M. Ali Aydin (1) , Ahmet Sertbas (1) , Tülin Atmaca (2, 3, 4, 5)

1 Department of Computer Engineering [Istanbul]
2 METHODES-SAMOVAR - Méthodes et modèles pour les réseaux
3 TSP - RST - Département Réseaux et Services de Télécommunications
4 SAMOVAR - Services répartis, Architectures, MOdélisation, Validation, Administration des Réseaux
5 CNRS - Centre National de la Recherche Scientifique

In this study, performance of classification algorithms N-Nearest Neighbors (N3) and Binned Nearest Neighbor (BNN) are analyzed in terms of Indoor localizations. Fingerprint method which is based on Received Signal Strength Indication (RSSI) is taken into consideration. RSSI is a measurement of the power present in a received radio signal from transmitter. in this methode, the RSSI information is catured at the reference points and recorded for creating a signal map. the obtained signal map is knows as fingerprint signal map and in the second stage of algorithm is creating a positioning model to detect individual's position with the help of fingerprint signal map. In this work, N-Nearest Neighbors (N3) and Binned Nearest Neighbor(BNN)algorithms are used to create an indoor positioning model. For this purpose; two different signal maps are used to test the algorithms. UJIIndoorLoc includes multi-building and multi floor signal information while different from this RFKON includes a single-building single floor signal Information. N-nearest Neighbors (N3)and Binned Nearest Neighbor(BNN)algorithms are presented comparatively with respect to success of finding user position

Mots clés: Indoor localization; RSSI; Fingerprint; Accuracy; N3 and BNN algorithms

Domaines: Réseaux et télécommunications [cs.NI]

Serpil Ustebay, M. Ali Aydin, Ahmet Sertbas, Tülin Atmaca. A comparative analysis of N-Nearest Neighbors (N3) and Binned Nearest Neighbors (BNN) algorithms for indoor localization. CN 2017 : 24th international conference on Computer Networks, Jun 2017, Lądek Zdrój, Poland. pp.81 - 90, ⟨10.1007/978-3-319-59767-6_7⟩. ⟨hal-01691760⟩