Máquina de vetor de relevância aplicada à estimação do canal multipercurso faixa-larga

Abstract

The knowledge of the characteristics of the channel is fundamental to the design of high performance wireless communication systems. Herein is described the design of a wideband channel sounder, intended to supply experimental data regarding the propagation conditions of the signal over the scenario under study. The proper interpretation of the data acquired in the sounding stage plays an important role in the construction of channel models, and the use of methods that perform this task in accurate and reliable means becomes important. The implemented solution makes use of the Relevance Vector Machine (RVM) algorithm, that consists of a regression method based on Bayesian Statistics, to identify the constitutive parameters of the channel impulse response. The outputs of the estimation algorithm enables the construction of statistical models that properly represent the studied environment