---

A Nonparametric Approach to Monitoring Latency in Fiber Optic Networks: The Mixed DEWMA-CUSUM Sign Chart

Andi Tenriola Bangsawan, Georgina Maria Tinungki*, Erna Tri Herdiani

---

Download PDF

---

Abstract:

The demand for reliable and fast data connections continues to rise alongside advancements in information technology, making fiber-optic networks the primary solution. However, maintaining network performance poses significant challenges, particularly in detecting latency anomalies caused by various factors. Conventional statistical monitoring approaches often rely on the assumption of normal distribution, which does not always align with characteristic of real-world network data. This study proposes the Mixed DEWMA-CUSUM Sign (MDCS) control chart, a nonparametric statistical method that combines DEWMA's sensitivity to small changes and CUSUM's efficiency in detecting process shifts. By employing a Sign-based approach, this method addresses the limitations of parametric techniques, ensuring more reliable performance in monitoring latency anomalies. The findings indicate that the MDCS method not only significantly improves anomaly detection but also contributes to the development of statistical methods for monitoring data with non-normal distributions, thereby enhancing overall network reliability.

Keywords: latency, fiber-optic network monitoring, mixed DEWMA-CUSUM sign chart, nonparametric statistical process control, data non-normal distribution