Description
As the gap widens between the number of security threats and the number of security professionals, the need for automated security tools becomes increasingly important. These automated systems assist security professionals by identifying and/or fixing potential vulnerabilities before they can

As the gap widens between the number of security threats and the number of security professionals, the need for automated security tools becomes increasingly important. These automated systems assist security professionals by identifying and/or fixing potential vulnerabilities before they can be exploited. One such category of tools is exploit generators, which craft exploits to demonstrate a vulnerability and provide guidance on how to repair it. Existing exploit generators largely use the application code, either through static or dynamic analysis, to locate crashes and craft a payload.

This thesis proposes the Automated Reflection of CTF Hostile Exploits (ARCHES), an exploit generator that learns by example. ARCHES uses an inductive programming library named IRE to generate exploits from exploit examples. In doing so, ARCHES can create an exploit only from example exploit payloads without interacting with the service. By representing each component of the exploit interaction as a collection of theories for how that component occurs, ARCHES can identify critical state information and replicate an executable exploit. This methodology learns rapidly and works with only a few examples. The ARCHES exploit generator is targeted towards Capture the Flag (CTF) events as a suitable environment for initial research.

The effectiveness of this methodology was evaluated on four exploits with features that demonstrate the capabilities and limitations of this methodology. ARCHES is capable of reproducing exploits that require an understanding of state dependent input, such as a flag id. Additionally, ARCHES can handle basic utilization of state information that is revealed through service output. However, limitations in this methodology result in failure to replicate exploits that require a loop, intricate mathematics, or multiple TCP connections.

Inductive programming has potential as a security tool to augment existing automated security tools. Future research into these techniques will provide more capabilities for security professionals in academia and in industry.
Downloads
pdf (332 KB)

Details

Title
  • Automated reflection of CTF hostile exploits (ARCHES): inductive programming techniques for network traffic comprehension and reflection
Contributors
Date Created
2019
Resource Type
  • Text
  • Collections this item is in
    Note
    • Partial requirement for: M.S., Arizona State University, 2019
      Note type
      thesis
    • Includes bibliographical references (pages 47-49)
      Note type
      bibliography
    • Field of study: Computer science

    Citation and reuse

    Statement of Responsibility

    by Zackary Crosley

    Machine-readable links