Publication: Localización distribuida en redes de sensores binarios
Loading...
Identifiers
Publication date
2011-02
Defense date
2011-02-22
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
El presente Proyecto Fin de Carrera desarrolla un estudio sobre el problema de estimar la localización de un agente por medio de redes de sensores que emplean técnicas distribuidas. Concretamente, se analiza la influencia de diversos criterios de diseño de la red en la precisión de dicho problema, mediante el uso de tres reglas de fusión diferentes: estimador de máxima verosimilitud, estimador de mínimo riesgo empírico y un tercer método propuesto en este proyecto basado en distancias. Para ello, se emplea una red homogénea de sensores binarios cuya información consiste en decidir si han detectado o no la presencia del agente en la que la probabilidad de detección de cada sensor, pd, se modela como una función de la distancia entre el sensor y el agente. Además, para mejorar los recursos energéticos de la red se analiza el rendimiento obtenido por un esquema de transmisión en el que sólo los sensores con lecturas positivas transmiten su información. Por medio de la simulación de los distintos estimadores en diferentes escenarios se logra obtener medidas cuantitativas de precisión del proceso de localización. _____________________________________________________________________________________________________________
The present Master Thesis develops a study of target location estimation problem in sensor networks by using distributed techniques. Speci cally, the in fluence of several network design criteria in the accuracy of that problem is analyzed. This analysis is carried out by means of three different fusion rules to estimate the target location: a maximum likelihood estimation, a minimum empirical risk estimation and a third method, proposed in this project, based on distances. A binary homogeneous sensor network whose information consists in deciding if the target is detected or not is used and the probability of detection of each sensor in the network, pd, is modeled as a function of the distance between sensor and target. Moreover, in order to improve the energy resources of the network, it is studied the performance of a transmission scheme where only sensors with a positive detection transmit their information. By simulating the different estimation methods in several environments, quantitative measures of target location accuracy are achieved.
The present Master Thesis develops a study of target location estimation problem in sensor networks by using distributed techniques. Speci cally, the in fluence of several network design criteria in the accuracy of that problem is analyzed. This analysis is carried out by means of three different fusion rules to estimate the target location: a maximum likelihood estimation, a minimum empirical risk estimation and a third method, proposed in this project, based on distances. A binary homogeneous sensor network whose information consists in deciding if the target is detected or not is used and the probability of detection of each sensor in the network, pd, is modeled as a function of the distance between sensor and target. Moreover, in order to improve the energy resources of the network, it is studied the performance of a transmission scheme where only sensors with a positive detection transmit their information. By simulating the different estimation methods in several environments, quantitative measures of target location accuracy are achieved.
Description
Keywords
Redes de sensores, Radiocomunicación