REPUBLIC OF SERBIA MINISTRY OF DEFENCE
MINISTRY OF DEFENCE Material Resources Sector Defensive Technologies Department
|
COMPARATIVE ANALYSiS OF KNN, MLP AND GRNN WLAN INDOOR BASED POSITIONING TECHNIQUES
Faculty of Electrical Engineering, Banja Luka, nebojsa.maletic@etfbl.net Faculty of Electrical Engineering, Banja Luka, Faculty of Electrical Engineering, Banja Luka, Milan šunjevarić Institute for Computer Based Systems, RT-RK, Novi Sad, micosun@eunet.rs
Abstract: Indoor positioning systems relying on Wireless Local Area Network (WLAN) have become very popular in the last decade. Increased WLAN deployment and its wide infrastructure availability provide wide range of applications and services that can be offered to the end user. Thus, accurate and reliable positioning system is prerequisite aiming to meet given purpose. Three location estimation schemes: K Nearest Neighbors (KNN), Multilayer Perceptron (MLP), and Generalized Regression Neural Network (GRNN) are analyzed in this paper. These schemes are based on scene analysis. The office environment at the first floor of the Faculty of Electrical Engineering with the existing Extended Service Set is used as a test bed. The performance of these methods in terms of accuracy and precision is evaluated, comparative analysis is given and some conclusions are drawn. Key words: Indoor positioning, WLAN, KNN, MLP, GRNN.
|
|||||||||||
|