DESIGN AND PERFORMANCE OPTIMIZATION OF CAMPUS WI-FI NETWORK (CASE STUDY OF UNIZIK)

Subject Area: COMMUNICATION SYSTEM ENGINEERING

A Wi-Fi network is a wireless local area network designed to provide internet connectivity to people within a particular area. In the context of this study, the University of Nnamdi Azikiwe (UNIZIK) Wi-Fi Campus was considered the network of choice for optimization. To solve this problem, three years of data, which model the network at congestion and normal conditions, were collected and then used to train a Multi-Layered Neural Network (MLNN) algorithm and generate a model for real-time tracking and detection of congestion. The methodology used for this project began with a technical investigation of the quality-of-service performance at the case study campus network through collection of data. The collected data were presents and analyzed to determine the impact of user and application diversity on quality of service. Secondly, a machine learning algorithm was trained with data of dynamic congestion collected from UNIZIK campus network to develop a model for the detection of dynamic congestion. The model generated was integrated with a congestion management system and used to improve quality of service optimization at UNIZIK, using Simulink and classification learner software. The new model was integrated into the campus network and evaluated for 2 hours, with data collected at a 5-minute interval. The result, when evaluated considering throughput, reported 83.4% In conclusion, the integration of MLNN with Early Warning Systems(EWS) has demonstrated success in optimizing the campus network. Recommendations to extend the new congestion solution to other network types apart from campus networks were suggested.