Cooperative Localization in Mines Using Fingerprinting and Neural Networks

Dayekh, Shehadi (2010). Cooperative Localization in Mines Using Fingerprinting and Neural Networks. (Mémoire de maîtrise). Université du Québec en Abitibi-Témiscamingue. Repéré dans Depositum à https://depositum.uqat.ca/id/eprint/54

Prévisualisation

PDF Télécharger (7MB) | Prévisualisation

Résumé

This work is a special investigation in the localization of users in underground and confined areas such as gold mines. It sheds light on the basic approaches that are used nowadays to estimate the position and track users using wireless technology. Localization or Geo-location in confined and underground areas is one of the topics under research in mining labs and industries. The position of personnel and equipments in areas such as mines is of high importance because it improves industrial safety and security. Due to the special nature of underground environments, signals transmitted in a mine gallery suffer severe multipath effects caused by reflection, refraction, diffraction and collision with humid rough surfaces. In such cases and in cases where the signals are blocked due to the non-line of sight (NLOS) regions, traditional localization techniques based on the RSS, AOA and TOA/TDOA lead to high position estimation errors. One of the proposed solutions to such challenging situations is based on extracting the channel impulse response fingerprints with reference to one wireless receiver and using an artificial neural network as the matching algorithm.
In this work we study this approach in a multiple access network where multiple access points are present. The diversity of the collected fingerprints allows us to create artificial neural networks that work separately or cooperatively using the same localization technique. In this approach, the received signals by the mobile at various distances are analysed and several components of each signal are extracted accordingly. The channel impulse response found at each position is unique to the position of the receiver. The parameters extracted from the CIR are the received signal strength, mean excess delay, root mean square, maximum excess delay, the number of multipath components, the total power of the received signal, the power of the first arrival and the delay of the first arrival.
The use of multiple fingerprints from multiple references not only adds diversity to the set of inputs fed to the neural network but it also enhances the overall concept and makes it applicable in a multi-access environment. Localization is analyzed in the presence of two receivers using several position estimation procedures. The results showed that using two CIRs in a cooperative localization technique gives a position accuracy less than or equal to 1m for 90% of both trained and untrained neural networks. Another way of using cooperative intelligence is by using the time domain including tracking, probabilities and previous positions to the localization system. Estimating new positions based on previous positions recorded in history has a great improvement factor on the accuracy of the localization system where it showed an estimation error of less than 50cm for 90% of training data and 65cm for testing data.
The details of those techniques and the estimation errors and graphs are fully presented and they show that using cooperative artificial intelligence in the presence of multiple signatures from different reference points as well as using tracking improves significantly the accuracy, precision, scalability and the overall performance of the localization system.

Type de document:	Thèse ou mémoires (Mémoire de maîtrise)
Directeur de mémoire/thèse:	Kandil, Nahi
Codirecteurs de mémoire/thèse:	Affes , Sofiène et Nerguizian, Chahé
Informations complémentaires:	Comprend un résumé. Mémoire de maîtrise, soumis à l'école d'ingénierie, Département des sciences appliquées comme exigence partielle de la maîtrise en ingénierie (option Télécommunications) Comprend des réf. bibliogr. (f. 61-64)
Mots-clés libres:	empreinte digital mine reseau neuronal localisation ingenierie genie telecommunication espace confine souterrain signal sans fil positionnement
Divisions:	Génie > Maîtrise en ingénierie
Date de dépôt:	16 déc. 2011 18:44
Dernière modification:	21 févr. 2013 20:45
URI:	https://depositum.uqat.ca/id/eprint/54

Gestion Actions (Identification requise)

Dernière vérification avant le dépôt