Adversarial detection games in network security applications with imperfect and incomplete information

Parras Moral, Juan (2020). Adversarial detection games in network security applications with imperfect and incomplete information. Thesis (Doctoral), E.T.S.I. Telecomunicación (UPM). https://doi.org/10.20868/UPM.thesis.57986.

Description

Title: Adversarial detection games in network security applications with imperfect and incomplete information
Author/s:
  • Parras Moral, Juan
Contributor/s:
  • Zazo Bello, Santiago
Item Type: Thesis (Doctoral)
Date: 2020
Subjects:
Faculty: E.T.S.I. Telecomunicación (UPM)
Department: Señales, Sistemas y Radiocomunicaciones
Creative Commons Licenses: Recognition - No derivative works - Non commercial

Full text

[img]
Preview
PDF - Requires a PDF viewer, such as GSview, Xpdf or Adobe Acrobat Reader
Download (3MB) | Preview

Abstract

This Ph.D. thesis deals with security problems in Wireless Sensor Networks. As the number of devices interconnected grows, the amount of threats and vulnerabilities also increases. Namely, in this thesis, we focus on two family of attacks: the backoff attack, which affects to the multiple access to a shared wireless channel, and the spectrum sensing data falsification attack, which arises in networks which try to make a decision about the state of a spectrum channel cooperatively. First, we use game theory tools to model the backoff attacks. We start by introducing two different algorithms that can be used to learn in discounted repeated games. Then, we motivate the importance of the backoff attack by showing analytically its effects on the network resources, which are not shared evenly as the attacking sensors receive a larger part of the network throughput. Afterwards, we show that the backoff attack can be modeled, under certain assumptions, using game theory tools, namely, static and repeated games, and provide analytical solutions and also algorithms to learn these solutions. A problem that arises for the defense mechanism is that it is possible that the agent is able to adapt to it. We then explore what happens if the agent knows the defense mechanism and acts in such a way that it is able to exploit the defense mechanism without being discovered. As we show, this is a significant threat to both attacks studied in this work, as the agent is able to successfully exploit the defense mechanism: in order to alleviate this attack, we propose a novel detection framework that is successful against such attack. However, we can even develop attack strategies that do not need the agent to know the defense mechanism: by means of reinforcement learning tools, it is able even to exploit a possibly unknown mechanism simply by interacting with it. Hence, these attack strategies are a significant threat against current defense mechanisms. We finally develop a defense mechanism against such intelligent attackers, based on inverse reinforcement learning tools, which is able to successfully mitigate the attack effects. ----------RESUMEN---------- Esta tesis trata con problemas de seguridad en redes de sensores inalámbricas. Debido a que el número de dispositivos interconectados crece, la cantidad de amenazas y vulnerabilidades también lo hace. Concretamente, en esta tesis nos centramos en dos familias de ataques: los ataques de backoff, que afectan al acceso múltiple a un canal inalámbrico compartido, y el ataque de falsificación de datos de sensado de espectro, que surge en redes que tratan de tomar una decisión respecto al estado del canal de manera cooperativa. En primer lugar, usamos herramientas de teoría de juegos para modelar el ataque de backoff. Comenzamos introduciendo dos algoritmos diferentes que se pueden usar para aprender juegos repetidos con descuento. Luego, desarrollamos la importancia del ataque de backoff al mostrar de forma analítica sus efectos sobre los recursos de la red, ya que provoca que estos no sean distribuidos de manera uniforme, debido a que los sensores atacantes obtienen una porción mayor del ancho de banda de la red. Después, mostramos que el ataque de backoff puede ser modelado, bajo ciertas asunciones, usando herramientas de teoría de juegos; concretamente, juegos estáticos y repetidos, y proporcionamos soluciones analíticas y algoritmos que aprenden estas soluciones. Un problema para el mecanismo de defensa es que es posible que el agente se pueda adaptar. Así que pasamos a explorar que ocurre si el agente conoce el mecanismo de defensa y actúa de tal manera que es capaz de atacarlo sin ser descubierto. Como mostramos, esta es una amenaza significativa para ambos ataques estudiados en este trabajo, ya que el agente es capaz de tener éxito burlando al sistema de defensa. Para mitigar los efectos de este ataque, proponemos un modelo novedoso de detección que tiene éxito frente a estos ataques. Sin embargo, podemos incluso desarrollar estrategias de ataque que no requieren que el agente conozca el mecanismo de defensa: usando herramientas de aprendizaje por refuerzo, el atacante es capaz de atacar un mecanismo de defensa posiblemente desconocido simplemente interactuando con él. De modo que estas estrategias son una amenaza significativa contra los sistemas de defensa actuales. Finalmente, desarrollamos un mecanismo de defensa contra estos ataques inteligentes, basado en aprendizaje por refuerzo inverso, que es capaz de mitigar con éxito los efectos del ataque.

More information

Item ID: 57986
DC Identifier: http://oa.upm.es/57986/
OAI Identifier: oai:oa.upm.es:57986
DOI: 10.20868/UPM.thesis.57986
Deposited by: Archivo Digital UPM 2
Deposited on: 17 Feb 2020 07:25
Last Modified: 17 Aug 2020 22:30
  • Logo InvestigaM (UPM)
  • Logo GEOUP4
  • Logo Open Access
  • Open Access
  • Logo Sherpa/Romeo
    Check whether the anglo-saxon journal in which you have published an article allows you to also publish it under open access.
  • Logo Dulcinea
    Check whether the spanish journal in which you have published an article allows you to also publish it under open access.
  • Logo de Recolecta
  • Logo del Observatorio I+D+i UPM
  • Logo de OpenCourseWare UPM