Anomaly Detection Using Call Stack Information

  • Authors:
  • Henry Hanping Feng;Oleg M. Kolesnikov;Prahlad Fogla;Wenke Lee;Weibo Gong

  • Affiliations:
  • -;-;-;-;-

  • Venue:
  • SP '03 Proceedings of the 2003 IEEE Symposium on Security and Privacy
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

The call stack of a program execution can be a very goodinformation source for intrusion detection. There is no priorwork on dynamically extracting information from call stackand effectively using it to detect exploits. In this paper, wepropose a new method to do anomaly detection using callstack information. The basic idea is to extract return addressesfrom the call stack, and generate abstract executionpath between two program execution points. Experimentsshow that our method can detect some attacks thatcannot be detected by other approaches, while its convergenceand false positive performance is comparable to orbetter than the other approaches. We compare our methodwith other approaches by analyzing their underlying principlesand thus achieve a better characterization of theirperformance, in particular, on what and why attacks will bemissed by the various approaches.