Modelagem Neuro-Fuzzy de perdas de propagação para planejamento de redes LTE

Abstract

The development of mobile communication technologies is associated to the demand of services by users who are more and more using the data services than voice. And to keep up with those demands the telecommunication enterprises seek to meet with new technologies this user searching for speed and quality in the service used by mobile networks. The technology LTE (Long Term Evolution) has shown flexible features that stand out in relation to the technologies that preceded it as the GSM and UMTS and the operators to meet the goals of existing networks such as multimedia services, spectral efficiency and business for mobile broadband. The organization that developed the LTE specifications is the 3GPP (3rd Generation Partners Project) and standardized by the European Standards Institute in the area of Telecommunications ETSI (European Telecommunications Standard Institute). This dissertation was realized a study about propagation losses in outdoor environment, from models found in the literature and suitable for the propagation channel. Performance results are presented using the metrics of root mean square error and standard deviation and the graphical representation of data measured. And to helpful to understand this analysis is interesting to map the area and deploy or adapt planning tools of coverage with more accurate and efficient. The data used in this study was collected on some main roads in the Castanhal city, of Pará state, northern Brazil. The frequency of the transmitted signal used for LTE is 1.800 MHz and the method used to predict propagation loss was Neuro-Fuzzy. This system uses the techniques of Computational Intelligence that combines Artificial Neural Networks (ANN) and Fuzzy Logic (LF) and have demonstrated ability to solve different types of problems in various applications in different areas. And through this technique, the comparison between the results predicted by the proposed method and literature propagation models to provide an analysis by the signal characteristics in specific channels, observing the limitations and local features.