Honeypot back-propagation for mitigating spoofing distributed Denial-of-Service attacks

  • Authors:
  • Sherif Khattab;Rami Melhem;Daniel Mossé;Taieb Znati

  • Affiliations:
  • Department of Computer Science, University of Pittsburgh, PA;Department of Computer Science, University of Pittsburgh, PA;Department of Computer Science, University of Pittsburgh, PA;Department of Computer Science, University of Pittsburgh, PA and Department of Information Science and Telecommunications, University of Pittsburgh, PA

  • Venue:
  • Journal of Parallel and Distributed Computing - Special issue: Security in grid and distributed systems
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

The Denial-of-Service (DoS) attack is a challenging problem in the current Internet. Many schemes have been proposed to trace spoofed (forged) attack packets back to their sources. Among them, hop-by-hop schemes are less vulnerable to router compromise than packet marking schemes, but they require accurate attack signatures, high storage or bandwidth overhead, and cooperation of many ISPs. In this paper, we propose honeypot back-propagation, an efficient hop-by-hop traceback mechanism, in which accurate attack signatures are obtained by a novel leverage of the roaming honeypots scheme. The reception of attack packets by a roaming honeypot (a decoy machine camouflaged within a server pool) triggers the activation of a tree of honeypot sessions rooted at the honeypot under attack toward attack sources. The tree is formed hierarchically, first at Autonomous system (AS) level and then at router level. Honeypot back-propagation supports incremental deployment by providing incentives for ISPs even with partial deployment. Against low-rate attackers, most traceback schemes would take a long time to collect the needed number of packets. To address this problem, we also propose progressive back-propagation to handle low-rate attacks, such as on-off attacks with short bursts. Analytical and simulation results demonstrate the effectiveness of the proposed schemes under a variety of DDoS attack scenarios.