Redes neurais artificiais aplicadas ao problema da localização em ambientes fechados

AUTOR(ES)
DATA DE PUBLICAÇÃO

2009

RESUMO

This thesis addresses the indoor location problem using on artificial neural networks-based techniques. In this system, the received signal strength information (RSSI) provided by standard network wireless interfaces are the basis for mobile devices location prediction. Traditional methods of indoor location have several undesirable characteristics, such as implementation difficulties, lack of flexibility (requiring APs specific position), high number of parameters and high computational cost. Traditional indoor location algorithms such as the Nearest Neighbor Algorithm were compared to methods based on Multilayer Neural Networks (Perceptron MLP) and the Kohonen self-organized map. We conclude that the Kohonens implementation is able to provide significantly better results (less errors, faster localization) than those obtained in recent studies of indoor localization.

ASSUNTO(S)

engenharia elétrica redes neurais (computação) engenharia eletrica

ACESSO AO ARTIGO

Documentos Relacionados