Skip to Main content Skip to Navigation

Application des techniques d'apprentissage à la géolocalisation par radio fingerprint

Abstract : The objective of this thesis is to improve the localization precision in indoor environments where GPS signal is weak or non‐existent. The originality of the research lies in the design of indoor localization system based on radio signals received from external sources. In this work, real measurements of received GSM signals were used. Given the difficulties associated with these experimental procedures (uncertainty and noise related to the measurements), the first part of the thesis is dedicated to their description. The processing and interpretation of radio GSM measurements in a real propagation environment were a challenge in this study. Indeed, all carriers of the GSM network were considered with no a priori hypothesis made regarding their relevance. The second part of this thesis then describes a study of relevance, as well as the statistical learning approaches (supervised and semi‐supervised) which were implemented to predict positions from the measurements. A choice of learning method was made based on the studies conducted on these measurements. The promising room‐level localization performance reached in this thesis demonstrates clearly that good quality indoor localization can be obtained by applying a machine learning strategy to GSM radio fingerprints.
Complete list of metadatas

Cited literature [82 references]  Display  Hide  Download
Contributor : Daniel Swift <>
Submitted on : Wednesday, December 15, 2010 - 11:14:34 AM
Last modification on : Wednesday, October 14, 2020 - 3:43:08 AM
Long-term archiving on: : Friday, December 2, 2016 - 10:35:47 PM


  • HAL Id : pastel-00546952, version 1


Iness Ahriz Roula. Application des techniques d'apprentissage à la géolocalisation par radio fingerprint. Analyse de données, Statistiques et Probabilités []. Université Pierre et Marie Curie - Paris VI, 2010. Français. ⟨pastel-00546952⟩



Record views


Files downloads