An enhanced localisation system for indoor environment using clustering technique. (2017)

• Record Type: Journal Article
• Authors: Alhammadi, Abdulraqeb; Alias, Mohamad Yusoff; Tan, Su-Wei; Sapumohotti, Chamal
• Is Part Of: International journal of computational vision and robotics. Volume 7:Number 1/2(2017)
• Page Start: 83
• Page End: 98
• Publication Date: 2017
• Subjects: Bayesian networks -- deterministic approach -- K-nearest neighbour -- kNN -- probabilistic approach -- RF fingerprinting -- clustering -- localisation systems -- indoor environments -- wireless LANs -- local area networks -- WLAN -- mobile devices; Computer vision -- Periodicals; Robotics -- Periodicals; Artificial intelligence -- Periodicals; 006.3705
• Languages: English
• ISSNs: 1752-9131

Abstract

Recently, indoor localisation techniques that use wireless local area network (WLAN) be a-con signals have gained much attention by the research communities. Many localisations methods are used to estimate the user of mobile device in indoor environments. However, the accuracy of these methods is affected by the nature of the test-bed environment. In this paper, we introduce an experimental test-bed in a typical indoor environment. We used a finger printing-based localisation algorithm to estimate the user location. The fingerprinting technique consists of two phases: offline phase and online phase. In the offline phase, calibration points are collected at certain places in floor to build a radio map. In the online phase, deterministic and probabilistic approaches are applied in order to get the correct estimated position of a mobile device. In deterministic approach, the position of mobile device estimated by K-nearest neighbour (KNN). In probabilistic approach, the position of mobile device estimated by Bayesian network (BN). Clustering technique is proposed to improve the system's accuracy and reduce the radio map size in the offline phase. We present experimental results that improved the system accuracy and reduce the size of radio map by using the proposed clustering technique.