Nonlinear Analysis for Automation Purposes in Computer Networks
Nonlinear Analysis for Automation Purposes in Computer Networks

Author(s): Milan Milivojević, Milan Pavlović, Marija Zajeganović
Subject(s): Social Sciences
Published by: Udruženje ekonomista i menadžera Balkana
Keywords: Automation; Computer networks; Fractal analysis; Nonlinear statistical analysis; Python; Round-trip time
Summary/Abstract: Automated network management processes have become more common in recent years. The main goal of automation is to efficiently configure a computer network. This is especially true when software and hardware problems require a rapid response. In this context, selecting appropriate features to serve as inputs for machine learning algorithms is important. It is important to choose the most appropriate features that can help in the event of a network disruption, such as a component failure that requires network equipment to be reconfigured or emerging security threats. One of the key metrics that provides insight into the state of a computer network is round-trip time (RTT). The sequence of RTT data serves as a fundamental basis for extracting relevant features. While linear analysis provides basic features, the complex nature of these processes requires more advanced analytical methods. Nonlinear time series analysis naturally becomes necessary. Features extracted using these methods provide a much more accurate assessment of the network’s state. This paper presents the application of fractal analysis for feature extraction to assess network conditions and automate network configuration processes. Fractal analysis, a well-established nonlinear method, is widely recognized in the literature. This paper explores the potential of applying fractal analysis to time series containing RTT data. Simulations were performed using the GNS3 network simulator and the Python programming language.

• Details
• Contents