2010 | OriginalPaper | Buchkapitel
Modeling and Containment of Search Worms Targeting Web Applications
verfasst von : Jingyu Hua, Kouichi Sakurai
Erschienen in: Detection of Intrusions and Malware, and Vulnerability Assessment
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Many web applications leak sensitive pages (we name them
eigenpages
) that can disclose their vulnerabilities. As a result, some worms like
Santy
locate their targets by searching specific eigenpages in search engines with well-crafted keywords. Such worms are so called
search worms
. In this paper, we focus on the modeling and containment of these search worms. We first study the influence of the eigenpage distribution on their spreading by introducing two propagation models: U-Model assuming eigenpages uniformly distributed on servers and PL-Model assuming the distribution follows a power law. We show that the uniform distribution maximizes the spreading speed of the search worm. Then we study the influence of the page ranking and introduce another propagation model: PR-Model. In this model, search results are ranked based on their PageRank values and the relative importance of their resident servers. Finally, we propose a containment system for search worms based on honey-page insertion: a small number of fake pages which will induce visitors to pre-established honeypots are randomly inserted into search results, and then infectious can be detected and reported to search engines when their malicious scans hit honeypots. We study the relationship between the containment effectiveness and the honey-page insert rate with our propagation models and find that the Santy worm can be almost completely stopped at its early age by inserting no more than 2 honey pages in every 100 search results, which is extremely effective.