Artificial intelligence applications for foult root cause analysis in transport systems used in telecommunication systems

In this thesis study, artificial intelligence applications for root cause analysis in transport systems used in telecommunications are explained. Data obtained from these systems were analyzed using five different classification methods. Genetic Algorithms (GA) and Synthetic Minority Oversampling Technique (SMOTE) models were also employed in the study. The analysis includes data collected from Fiber Optic Cables, Dense Wavelength Division Multiplexing (DWDM) systems from the transmission layer, and devices from the Internet Protocol Multi-Protocol Label Switching (IP-MPLS) layer, which together form the transport layer of telecommunications systems. Normally, these layers have independent data collection and management systems, but in this study, inventory matching was performed between them, and relevant parameters were selected using domain expertise. The selected features from the data pool were used for further analysis. The output of the data pool focused on three key issues that are significant problems in telecommunications systems, leading to performance errors or frequent service disruptions: "MPLS Flap," "Short-term Outage," and "Long-term Outage". In particular, the issue of establishing the relationship between different domains of expertise and identifying the root cause of the error by matching the errors in the layer close to the customer with the data in the deeper layers has produced a very important and useful output as it positively affects the quality of the service provided to the customer in telecommunication systems. As artificial intelligence classification methods in the study; Random Forest (RF), Decision Tree (DT), eXtreme Gradient Boosting (XGBoost), Naive Bayes, K-Nearest Neighbors (KNN) methods were used. When the results were examined, it was seen that the results of the Random Forest (RF) method made the most accurate estimation.