Unveiling Latency-Induced Service Degradation: A Methodological Approach with Dataset

Abstract

This paper presents a comprehensive study on the identification and analysis of Service Degradation (SD) events within a university dormitory network, leveraging LAN data to develop a robust methodology applicable to diverse networking environments. Employing statistical techniques, such as Interquartile Range (IQR) and Z-score analyses, we detect significant deviations in network performance - specifically, extreme delays and jitter - that indicate potential SD. The methodology was rigorously validated in various settings, demonstrating minimal deviations in results and reinforcing the approach's consistency and reliability. Initial tests conducted in a university dormitory environment suggest the model's potential applicability in both residential and enterprise networks, thus broadening its utility. By refining the detection and understanding of SD indicators, this research contributes systematic methodological applications and a valuable annotated dataset to the field. This groundwork enables network administrators to enhance service quality preemptively, offering significant implications for future research and practical applications in network management.