Rede neural aplicada em técnica de sincronismo de símbolo em transmissões digitais

Abstract

Due to the growth and advancement of technologies related to digital communication, there is a need to pursue precision and excellence in all components of current systems. As a result, the application of a method to correct symbol synchronization issues in digital transmissions, which occur due to temperature variations in some system components, is imperative. This study aims to replace the classic interpolator proposed by Gardner in the block responsible for symbol synchronization recovery by implementing an artificial neural network, specifically the radial basis neural network. In summary, the study seeks to compare an alternative method with the classic one already applied in current systems. Using real data with various levels of noise and distortion, the results demonstrated significant convergence, especially for deviations of 1080 ppm. However, for smaller deviations, improvements in network training are required. This study highlights the potential of neural networks in the field of data transmission and suggests the exploration of enhancement strategies for practical applications in enterprise networks and telecommunications systems. This work offers an innovative perspective for enhancing the integrity of high-speed data transmissions.