European Symposium on Research in Computer Security

ESORICS 2012: Computer Security – ESORICS 2012 pp 145-162

A Probabilistic Framework for Localization of Attackers in MANETs

  • Massimiliano Albanese
  • Alessandra De Benedictis
  • Sushil Jajodia
  • Paulo Shakarian
Conference paper

DOI: 10.1007/978-3-642-33167-1_9

Volume 7459 of the book series Lecture Notes in Computer Science (LNCS)
Cite this paper as:
Albanese M., De Benedictis A., Jajodia S., Shakarian P. (2012) A Probabilistic Framework for Localization of Attackers in MANETs. In: Foresti S., Yung M., Martinelli F. (eds) Computer Security – ESORICS 2012. ESORICS 2012. Lecture Notes in Computer Science, vol 7459. Springer, Berlin, Heidelberg

Abstract

Mobile Ad Hoc Networks (MANETs) represent an attractive and cost effective solution for providing connectivity in areas where a fixed infrastructure is not available or not a viable option. However, given their wireless nature and the lack of a stable infrastructure, MANETs are susceptible to a wide range of attacks waged by malicious nodes physically located within the transmission range of legitimate nodes. Whilst most research has focused on methods for detecting attacks, we propose a novel probabilistic framework for estimating – independently of the type of attack – the physical location of attackers, based on the location of nodes that have detected malicious activity in their neighborhood. We assume that certain countermeasures can be deployed to capture or isolate malicious nodes, and they can provide feedback on whether an attacker is actually present in a target region. We are interested in (i) estimating the minimum number of countermeasures that need to be deployed to isolate all attackers, and (ii) finding the deployment that maximizes either the expected number of attackers in the target regions or the expected number of alerts explained by the solution, subject to a constraint on the number of countermeasures. We show that these problems are NP-hard, and propose two polynomial time heuristic algorithms to find approximate solutions. The feedback provided by deployed countermeasures is taken into account to iteratively re-deploy them until all attackers are captured. Experiments using the network simulator NS-2 show that our approach works well in practice, and both algorithms can capture over 80% of the attackers within a few deployment cycles.

Keywords

Attacker localization MANET probabilistic framework 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  • Massimiliano Albanese
    • 1
  • Alessandra De Benedictis
    • 2
  • Sushil Jajodia
    • 1
  • Paulo Shakarian
    • 3
  1. 1.Center for Secure Information SystemsGeorge Mason UniversityFairfaxUSA
  2. 2.Department of Computer ScienceUniversity of Naples “Federico II”NaplesItaly
  3. 3.Department of Electrical Engineering and Computer ScienceUnited States Military AcademyWest PointUSA