Browsing by Author "Oney, Mehmet Ugur"

Now showing 1 - 3 of 3

A Comparison of Neural Network Approaches for Network Intrusion Detection
(Springer international Publishing Ag, 2020) Oney, Mehmet Ugur; Peker, Serhat; Software Engineering; 06. School Of Engineering; 01. Atılım University
Nowadays, network intrusion detection is an important area of research in computer network security, and the use of artificial neural networks (ANNs) have become increasingly popular in this field. Despite this, the research concerning comparison of artificial neural network architectures in the network intrusion detection is a relatively insufficient. To make up for this lack, this study aims to examine the neural network architectures in network intrusion detection to determine which architecture performs best, and to examine the effects of the architectural components, such as optimization functions, activation functions, learning momentum on the performance. For this purpose, 6480 neural networks were generated, their performances were evaluated by conducting a series of experiments on KDD99 dataset, and the results were reported. This study will be a useful reference to researchers and practitioners hoping to use ANNs in network intrusion detection.
A Simulation Environment for Cybersecurity Attack Analysis Based on Network Traffic Logs
(inst Engineering Tech-iet, 2019) Daneshgadeh, Salva; Oney, Mehmet Ugur; Kemmerich, Thomas; Baykal, Nazife; 01. Atılım University
The continued and rapid progress of network technology has revolutionized all modern critical infrastructures and business models. Technologies today are firmly relying on network and communication facilities which in turn make them dependent on network security. Network-security investments do not always guarantee the security of organizations. However, the evaluation of security solutions requires designing, testing and developing sophisticated security tools which are often very expensive. Simulation and virtualization techniques empower researchers to adapt all experimental scenarios of network security in a more cost and time-effective manner before deciding about the final security solution. This study presents a detailed guideline to model and develop a simultaneous virtualized and simulated environment for computer networks to practice different network attack scenarios. The preliminary object of this study is to create a test bed for network anomaly detection research. The required dataset for anomaly or attack detection studies can be prepared based on the proposed environment in this study. We used open source GNS3 emulation tool, Docker containers, pfSense firewall, NTOPNG network traffic-monitoring tool, BoNeSi DDoS botnet simulator, Ostinato network workload generation tool and MYSQL database to collect simulated network traffic data. This simulation environment can also be utilized in a variety of cybersecurity studies such as vulnerability analysis, attack detection, penetration testing and monitoring by minor changes.
The Use of Artificial Neural Networks in Network Intrusion Detection: a Systematic Review
(Ieee, 2018) Oney, Mehmet Ugur; Peker, Serhat; Software Engineering; 06. School Of Engineering; 01. Atılım University
Network intrusion detection is an important research field and artificial neural networks have become increasingly popular in this subject. Despite this, there is a lack of systematic literature review on that issue. In this manner, the aim of this study to examine the studies concerning the application artificial neural network approaches in network intrusion detection to determine the general trends. For this purpose, the articles published within the last decade from 2008 to 2018 were systematically reviewed and 43 articles were retrieved from commonly used databases by using a search strategy. Then, these selected papers were classified by the publication type, the year of publication, the type of the neural network architectures they employed, and the dataset they used. The results indicate that there is a rising trend in the usage of ANN approaches in the network intrusion detection with the gaining popularity of deep neural networks in recent years. Moreover, the KDD'99 dataset is the most commonly used dataset in the studies of network intrusion detection using ANNs. We hope that this paper provides a roadmap to guide future research on network intrusion detection using ANNs.