On Effectiveness of Link Padding for Statistical Traffic Analysis Attacks

  • Authors:
  • Xinwen Fu;Bryan Graham;Riccardo Bettati;Wei Zhao

  • Affiliations:
  • -;-;-;-

  • Venue:
  • ICDCS '03 Proceedings of the 23rd International Conference on Distributed Computing Systems
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Traffic analysis attacks aim at deriving mission criticalinformation from the analysis of the traffic transmitted overa network. Countermeasures for such attacks are usuallyrealized by properly "padding" the payload traffic so thatthe statistics of the overall traffic become significantly differentfrom that of the payload traffic. In this paper, wepropose a analytical framework for traffic analysis attacksbased on statistical pattern recognition techniques. Westudy the effectiveness of countermeasures for traffic analysisattacks within our proposed framework. Two basic countermeasurestrategies are (a) to pad the traffic with constantinterarrival times of packets (CIT) or (b) to pad thetraffic with variable interarrival times (VIT). Our experimentsshow that CIT countermeasures fail when the adversaryuses sample variance or sample entropy of packet interarrivaltimes for statistical analysis. On the other hand,VIT countermeasures are effective regardless of which samplestatistics are used by the adversary. These observationsare validated by analysis of detection rates based on sampledistributions of packet interarrival times.