Particle filter approach for estimating indoor mobile user location using IEEE 802.11 signals (F1090)

• Main Authors: Mantoro, Teddy, Ayu, Media Anugerah, Raman, Shakiratul Husna, Md. Latiff , Nurul Hidayati
• Format: Conference or Workshop Item
• Language: English; English
• Published: 2011
• Subjects: T Technology (General)
• Online Access: http://irep.iium.edu.my/15025/1/Particle_filter_approach_for_estimating_indoor_mobile_user_location.pdf; http://irep.iium.edu.my/15025/4/Particle_Filter_Approach_for_Tracking_Indoor_%28full_text%29.pdf; http://irep.iium.edu.my/15025/; http://sparc.nfu.edu.tw/~iceti2011/download/ICETI2011_Proceedings.pdf

To increase the accuracy of Location-aware personal computing application, multi-observers of IEEE 802.11 (Wi-Fi) signals can be used to track indoor user location. Even-though Wi-Fi is more and more widely available on most mobile devices, unfortunately, because of the reflection, refraction, temperature, humidity and the dynamic changing in the environment, the reading of Wi-Fi’s signal fluctuates greatly; the deviation can reach up to 33% from single Wi-Fi’s access point. This creates problem in tracking user location indoor. Moreover, the use of light estimation algorithms such as fingerprinting, ranking algorithm, Weighted Centroid method, k-Nearest Neighbour, did not give a good tracking result. This paper proposes the use of Particle Filter in improving user location estimation which involves the modeling of non-linear and non-Gaussian systems. The aim is to increase the accuracy of tracking user location indoor. In our experiments, the real time data of multi-observer Wi-Fi signals have been used and the loss of diversity and parameter chosen in order to reduce the ambiguity has also been observed. We improve the algorithm in reducing the computational complexity by giving target/reference points. The paper discussed the comparison between the true location and the estimated location based on two types of signals data: normal data and noise data. The location estimation is predicted based on real-time signal and then compare it to the training data set. This approach shows a promising result in tracking user location indoor using particle filter algorithm.