Localização de sensores considerando custo mínimo

• Ano de defesa: 2013
• Autor(a) principal: Angelo Ferreira Assis
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Localização; Algoritmo genético; Redes de sensores sem fio; Computação; Algoritmos genéticos
• Link de acesso: http://hdl.handle.net/1843/ESBF-97CMA7

Localization is one of the key issues in Wireless Sensor Networks. Its use is undoubtedlyimportant in many different applications. However, its often required to minimize thelocalization cost in a network. This can be done by setting some nodes as anchors,which are used as reference to other nodes. Current literature solutions focus on findingas many nodes as possible given a static set of anchor nodes, providing each one of thesenodes resources such as a GPS to determine its location. On the other hand, this maybe unfeasible for many sensor networks due to the high cost and/or implementationcomplexity. The optimization problem presented in this work consists of finding theminimum set of anchor nodes needed to locate all nodes in the network. Anotherapproach is to find the shortest path between the anchors. This way a robot can beused to visit and set the exact position of the nodes that work as anchors, that is, it isonly necessary to minimize the path, since the localization cost is the robot fuel. Hereit is presented a model for the problem using Genetic Algorithms in order to createthis problem a better solution. Several tests were performed to show the effectivenessof the strategy based on the number of anchors required to locate the entire network.The results have shown that the genetic algorithm reached, on average, a 50%-bettersolution than the greedy algorithm, having a feasible runtime.