The FTTH (Fiber To The Home) market currently needs new network maintenance technologies that can, economically and effectively, cope with massive fiber plants. However, operating these networks requires adequate means for an effective monitoring cost. Especially for troubleshooting faults that are associated with the possibility of remote identification of fiber breaks, which may exist in the network. Optical Time Domain Reflectometry (OTDR) techniques are widely used in point-to-point optical network topologies. Nevertheless, it has major limitations in tree-structured PONs (Passive Optical Networks), where all different branches backscatter the light in just one conventional OTDR trace with combined signals arriving on the OLT (Optical Line Terminal) side. Furthermore, passive power splitters used in FTTH networks input large attenuation, impoverishing the reflected signal. This makes the identification of the very branch affected by the problem practically impossible, when considering conventional analyses. The use of constraint-based techniques have been applied in a large amount of applications for Engineering Design, where the duties imposed for graphics and equations constraints result in valued features to CAD/CAE software capabilities. Currently, it provides a faster decision making capacity for engineers. This work applies the constraint based approach along with a Differential Evolutionary Algorithm to separate the superimposed OTDR signals, after the splitters of a FTTH Passive Optical Networks. This research introduces a new set of algorithms performing a coupling to an Optical Network (ON) CAD Design with its correspondent OTDR measurement signal, considering its geographical distribution branches of different lengths after the splitter. Results of this work are presented in a FTTN (Fiber To The Node) prototype arrangement, using a 1:8 passive power splitter.