Prognostic methods leveraging Machine Learning (ML) require consistent and complete datasets to achieve accurate predictions. However, in the context of cable prognosis based on reflectometry, assembling such datasets is challenging due to the prolonged process of accumulating real condition monitoring data over the lifespan of cables and also the lack of such available datasets in the literature as far as we are informed. Hence, this paper proposes a solution for relevant data generation to simulate the degradation of cables during their lifetime. This initial step is highly important in our proposed cable prognosis methodology, which integrates distributed reflectometry sensors and ML techniques to estimate the remaining useful life of multi-branched wired networks. To do so, two use cases of cable aging, using model-driven and experimental approaches, are studied in this paper. As a conclusion, a hybrid approach will be considered based on accelerated aging experiments and simulations to set up a pertinent dataset and provide a deeper understanding of the cable degradation process.