Botnets: a heuristic-based detection framework

  • Authors:
  • Luís Mendonça;Henrique Santos

  • Affiliations:
  • University of Minho, Braga, Portugal;University of Minho, Braga, Portugal

  • Venue:
  • Proceedings of the Fifth International Conference on Security of Information and Networks
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Many different approaches have been used to target Internet security throughout time. It is now easy to realize the attackers' motivational shifts from the early days of lonely, proud-based, virus development to the recent eras of cooperative Internet cyber criminality where high profit and damage became a reality. Among the vast spectrum of tools available to perpetrate attacks on information networks, botnets are becoming more and more popular due to its scalability, attack power, return and cooperation capabilities. Evolved programming skills and sophisticated tools also come to scene allowing the appearance of corresponding more evolved malware. New, constantly appearing, malware developments in host infection, deployment, maintenance, control and dissimulation of bots keep changing the existing detection vectors creating the need for detection systems that go beyond signature-based detection approaches. In that way, research and implementation of anomaly-based botnet detection systems is fundamental to keep up with the continuously changing scenario of polymorphic botnets variants. This paper presents the research and tests made to define an effective set of traffic parameters capable of modeling both normal and abnormal activity of networks, focusing on botnet activity detection through anomalous and cooperative behavior. A detection framework prototype is also proposed and tested using real traffic collected in the University of Minho campi edge. In the end, the results of the test-runs executed on the developed detection framework are presented and discussed.