Modelado del BER mediante regresión no lineal en una red de acceso FTTH GPON

Abstract

Abstract: Ensuring quality in optical networks is crucial to guarantee high-quality service. In this context, the importance of precise modeling of the BER (Bit Error Rate) to ensure optimal performance is emphasized, which has significant implications for the design and optimization of these networks. The objective of this thesis is to achieve precision in BER modeling through non linear regression techniques. The process relies on using the OptiSystem software to simulate an FTTH GPON network and gather data on BER behavior at different wavelengths. These data are subsequently processed using Python software for non-linear regression modeling, encompassing techniques such as logarithmic, polynomial, exponential regression, and machine learning. Upon analyzing the results, it is determined that the exponential regression model best fits the BER data in this particular context, suggesting an exponential behavior of BER in FTTH GPON networks.