STM Journals Peer-reviewed research from Consortium e-Learning Network

Journal Of Network security

An Experimental Approach of Machine Learning Algorithms to Detect Botnet DDoS Attacks

  1. Voore Subrahmanyam
  2. Vinnakota Shivani
  3. Siraboina Prasunanju
  4. Akula Pranay
  5. Sanikommu Shreya Reddy

Abstract

Botnets are one of the threats in a network to hamper the quality of the network by disrupting theresources of the network. These Botnets can be controlled remotely by Botmaster. The Machine Learning Algorithms play a major role to detect and control the Botnets that cause to DDoS attacks, malwares and phishing attacks that are vulnerable to network resources. The DDoS attacks are most dangerous malware events that disrupt whole network. To solve DDoS attacks, various methods and algorithms are proposed. In this study, we proposed K-means Unsupervised Learning (USML) algorithm. In the proposed methodology, we conduct a practical approach, analyzing by ML algorithms i.e., K-means algorithms for the detecting Botnet DDoS attacks. For experimental analysis, we consider the UNBS-NB real-time datasets. In this approach, we compare K-means algorithms with Support Vector Machine (SVM), Artificial Neural network (ANN), Naive Bayes (NB) and Decision Tree (DT) for performance-based comparison. In results, we find that K-means (USML) is showing better performance than other machine learning algorithms.

Keywords

Support