The limits of global scanning worm detectors in the presence of background noise

  • Authors:
  • David W. Richardson;Steven D. Gribble;Edward D. Lazowska

  • Affiliations:
  • University of Washington;University of Washington;University of Washington

  • Venue:
  • Proceedings of the 2005 ACM workshop on Rapid malcode
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Internet worms cause billions of dollars in damage each year. To combat them, researchers have been exploring global worm detection systems to spot a new random scanning worm outbreak quickly. These systems passively listen for worm probes on unused IP addresses, looking for anomalous increases in probe traffic to distinguish the emergence of a new worm from background Internet noise.In this paper, we use analytic modeling, simulation, and measurement to understand how background noise impacts the detection ability of global scanning worm detectors. We investigate the relationship between the average background noise level, the number of IP addresses monitored, and the detection latency for two classes of global scanning worm detectors: scan packet-based and victims-based schemes. Our results show how worm detection latency degrades as a function of the background noise level. To compensate, global scanning worm detectors can increase the number of IP addresses that they monitor. However, given the growth trend of background noise levels, the number of IP addresses which must be monitored may quickly become unreasonable. Because of this, we conclude that global scanning worm detection schemes are unlikely to be competitive with local scanning and signature-based worm detection schemes.