Influence of Model and Traffic Pattern on Determining the Self-Similarity in IP Networks

Abstract

This study aimed to determine the applicability of using selected libraries of computing environment R to establish the coefficient of self-similarity. R environment is an analytical environment with rich functionality that is used in many research and practical works concerning data analysis and knowledge discovery. Such an issue is significant in the context of contemporary wide area computer networks and the emerging type of network infrastructure IoT. This originates directly from the new nature of IoT traffic, which also has a substantial impact on Industry 4.0. It provides built-in operations facilitating data processing. The Hurst coefficient is used to evaluate traffic behavior and analyze its character. The study analyzed two cases of IoT network traffic: high and low intensity. For different sizes of time windows, we dermined the statistical Hurst exponent and compared it with standard, smoothed, and Robinson methods. The accuracy of the methods used was evaluated. Spectral regression graphs were additionally generated for selected motion variants. The obtained results were verified by Higuchi and Aggvar methods.

A Comparative Analysis of Selected Predictive Algorithms in Control of Machine Processes

“…In [6,7], the authors propose the use of fractal and multifractal analysis mechanisms, which can help to discover the structure of the communication system, especially the traffic pattern and characteristics, in order to better understand the threats and detect anomalies in the network performance. In particular, the authors' work presents the use of fractal analysis in detecting threats and anomalies.…”