Modeling Self-similar Packet Flow Transformation to Deterministic Using Kulback-Leibler Divergence for Network Optimization

Abstract

The research focuses on the stability of deterministic packet flows in telecommunication networks, vital for high QoS applications. Yet, the challenge lies in evaluating performance due to the absence of suitable analytical models for self-similar network traffic. The study aims to optimally estimate the time interval between packets. A mathematical model is proposed to convert self-similar input streams into normally distributed output streams, explored through limit transitions to deterministic flows. Key features involve optimizing the Kulback-Leibler divergence for input and output flow distributions. A groundbreaking Lemma is proven, showing that the minimum divergence occurs when mathematical expectations of time intervals in both streams are equal. Optimal time intervals in deterministic flows align with the input’s expectation. These findings carry significant implications for enhancing telecommunication network performance, providing valuable insights into stable behavior and enabling better QoS for critical applications.