BotNet Detection for Network Traffic using Ensemble Machine Learning Method

Abstract

In todays era the need of security is raising due to hike in security risks discovered every day. A new vulnerability can be found in any software or product by the attacker as it launches in the market. Botnet carried out various attacks in distributed manner which results in extensive disruption of network activity through information and identity theft, email spamming, click fraud DDoS (Distributed Denial of Service) attacks, virtual deceit and distributed resource usage for cryptocurrency mining. The main aim f botnet is to steal private data of clients, sendind spam and viruses and DOS attacks in the network. The detection of Botnet like Rbot ,Virut and Neris are still vigorous research area due to unavailability of any technique to detect the entire ecosystem of botnet. As they are comprised of different configurations and profoundly armored by malwares writers to dodge detection systems by utilizing complicated dodging techniques. Hence only solution is to discover the infected botnets to control over the services and ports. This work aims to contribute in the botnet detection with its overview and existing methods. The study focuses on techniques like one-hot encoding and variance thresholding. These techniques are utilized to clean the botnet dataset. The performance of the machine learning model can be improved with feature selection methods. The work explores the dataset imbalance problem with the help of ensemble machine learning techniques. The performance is evaluated on the best received model that is trained and tested on datasets of various attacks.