Neural Network Implementation for Detection of Denial of Service Attacks

Denial of Service attacks are considered a major risk because they can easily interrupt a service, a business or educational process. These attacks are relatively simple to conduct, even by an unskilled attacker and cause significant loss. For this reason, it is particularly important that these attacks are detected, recognized and blocked in time. Most of the advanced methods and tools to protect against such attacks are based on monitoring and constant tracking in order to detect suspicious IP traffic. The application of these methods is associated with additional computational resources and expertise, which leads to subjectivity in the threat assessment. Therefore, it is necessary to propose methods for automated adaptive detection and recognition of Distributed Denial of Service attacks. This study presents a method for automated detection and recognition of some of the Distributed Denial of Service attacks, by means of an automated adaptive system, based on a multilayer neural network. It is trained both normal and with signals reflecting different traffic conditions when Distributed Denial of Service attacks occur. The neural network is tested to recognize baseline signals, representing different normal traffic conditions and to detect the abnormal traffic situations. The research is conducted for different kinds of internal Distributed Denial of Service attacks on a real Local Area Network. The obtained recognition accuracy results are represented and the achieved benefits are discussed. © 2020 School of Science, IHU. All Rights Reserved.