NWANKWO, KENECHUKWU CHRISTIAN (2026) DEVELOPMENT OF A NETWORK CONGESTION MANAGEMENT SYSTEM. Other thesis, Godfrey Okoye University, Enugu Nigeria.
|
Text
KENE.pdf Download (1MB) | Preview |
Abstract
Problems related to the occurrence of network congestion remain among the most important in the area of communication technologies infrastructure because they result in poor quality of services, increased delays and losses and decrease traffic capacity. The inability of reactive solutions of network congestion handling to solve effectively the occurring problems is evident when discussing dynamic networks. This research is devoted to proving how to develop and use a predictive system for detecting the state of network congestion based on XGBoost algorithm. The proposed approach relies on stack architecture where back-end is implemented using Python and Flask and the front-end is designed using React. To conduct the experiment the synthetic data set reflecting performance metrics for networks of different topologies has been used. Such indicators as utilization of bandwidth, packet loss ratio, latency, jitter, queue length and throughput were used to describe network performance. XGBoost classifier model has been built and used to design RESTful API that operates through Flask back-end. The React-based front end includes a simple dashboard that facilitates file upload of CSV format, prediction generation, visual analysis of the prediction results using donut chart representation, and report generation. In implementing this system, a Waterfall strategy was considered, while the system structure and behaviors were captured through UML diagramming techniques, using OOAD principles. According to the experimental results, the GBDT model provides predictions with high accuracy, hence outperforming the rest of the models. The system is, therefore, a viable machine learning tool in addressing the problem of network congestion.
| Item Type: | Thesis (Other) |
|---|---|
| Subjects: | Q Science > Q Science (General) |
| Divisions: | Faculty of Engineering, Science and Mathematics > School of Electronics and Computer Science |
| Depositing User: | Cynthia Ugwuoti |
| Date Deposited: | 22 Jul 2026 11:12 |
| Last Modified: | 22 Jul 2026 11:12 |
| URI: | http://eprints.gouni.edu.ng/id/eprint/5991 |
Actions (login required)
 |
View Item |
