TU Darmstadt / ULB / TUbiblio

Analyzing and Mitigating Security Threats in P2P Systems

Ismail, Hatem (2018):
Analyzing and Mitigating Security Threats in P2P Systems.
Darmstadt, Technische Universität, [Online-Edition: https://tuprints.ulb.tu-darmstadt.de/7812],
[Ph.D. Thesis]

Abstract

Peer-to-Peer (P2P) protocols increasingly underlie a growing diversity of net- worked applications (e.g., file sharing, streaming multimedia, storage, VoIP) especially as the decentralized P2P paradigm inherently fosters scalability and robustness. The growing application-oriented services also result in the evolu- tion of P2P systems spanning diverse data dissemination techniques, peer roles and topological structures. On the flip side, while decentralization and scalability are attractive, and common for all P2P systems, these design features also increase the P2P network’s exposure to a variety of security threats that can result in the degradation of services. In this thesis, we illustrate a set of important P2P attack types and subsequently develop approaches to secure P2P networks from these progressive and evolving attacks. Covering a comprehensive progression of P2P systems of increasing complexity (i.e., structured, unstructured and streaming), we evaluate the corresponding feasibility of conducting attacks and the resultant impact onto them. Subsequently, we investigate the progressive steps of detection, mitigation and sanitization potential to restore the requisite P2P functionality. Depending on the targeted P2P network model, we propose countermeasures that (a) are effective against a specific attack type and its possible variants, (b) are light-weight in execution, (c) are fully decentralized, i.e., do not depend on central entities, and (d) allow for both reactive and proactive mitigation. Our theoretical analysis and simulations demonstrate that our proposed attack detection/mitigation mechanisms can reach up to 90-100% detection accu- racy while inducing low overhead of 5-10% even when operating under severe attack scenarios.

Item Type: Ph.D. Thesis
Erschienen: 2018
Creators: Ismail, Hatem
Title: Analyzing and Mitigating Security Threats in P2P Systems
Language: English
Abstract:

Peer-to-Peer (P2P) protocols increasingly underlie a growing diversity of net- worked applications (e.g., file sharing, streaming multimedia, storage, VoIP) especially as the decentralized P2P paradigm inherently fosters scalability and robustness. The growing application-oriented services also result in the evolu- tion of P2P systems spanning diverse data dissemination techniques, peer roles and topological structures. On the flip side, while decentralization and scalability are attractive, and common for all P2P systems, these design features also increase the P2P network’s exposure to a variety of security threats that can result in the degradation of services. In this thesis, we illustrate a set of important P2P attack types and subsequently develop approaches to secure P2P networks from these progressive and evolving attacks. Covering a comprehensive progression of P2P systems of increasing complexity (i.e., structured, unstructured and streaming), we evaluate the corresponding feasibility of conducting attacks and the resultant impact onto them. Subsequently, we investigate the progressive steps of detection, mitigation and sanitization potential to restore the requisite P2P functionality. Depending on the targeted P2P network model, we propose countermeasures that (a) are effective against a specific attack type and its possible variants, (b) are light-weight in execution, (c) are fully decentralized, i.e., do not depend on central entities, and (d) allow for both reactive and proactive mitigation. Our theoretical analysis and simulations demonstrate that our proposed attack detection/mitigation mechanisms can reach up to 90-100% detection accu- racy while inducing low overhead of 5-10% even when operating under severe attack scenarios.

Place of Publication: Darmstadt
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Dependable Embedded Systems & Software
Date Deposited: 07 Oct 2018 19:55
Official URL: https://tuprints.ulb.tu-darmstadt.de/7812
URN: urn:nbn:de:tuda-tuprints-78124
Referees: Suri, Prof. Neeraj and Khelil, Prof. Abdelmajid
Refereed / Verteidigung / mdl. Prüfung: 13 September 2018
Alternative Abstract:
Alternative abstract Language
Peer-to-Peer (P2P) Netzwerke unterliegen zunehmend einer sich steigernden Diversität vernetzter Anwendungen (z.B. File Sharing, Streaming, Multimedia, Datenspeicherung, VoIP), insbesondere infolge der durch das dezentralisierte P2P-Paradigma begünstigten Skalierbarkeit und Robustheit. Die zunehmende Zahl anwendungsorientierter Dienste führt auch zu einer Evolution von P2PModellen, die verschiedenartige Techniken der Datenverbreitung, sowie verschiedene Rollen der Peers und topologische Strukturen, umfasst. Dezentralisierung und Skalierbarkeit sind attraktive Eigenschaften von allen P2P-Systemen, führen aber auch zu einer Anfälligkeit dieser Systeme gegenüber verschiedener Sicherheitsschwachstellen, die wiederum zu einer Minderung der Dienstgüte führen können. Die vorliegende Arbeit zeigt einen Satz wichtiger P2P-Angriffsarten auf und entwickelt Ansätze zum Schutz vor diesen sich abzeichnenden progressiven Angriffen. Wir evaluieren die Durchführbarkeit und Auswirkungen dieser Angriffe für P2P-Modelle steigender Komplexität (Structured, Unstructured und Streaming). Anschließend untersuchen wir aufeinander aufbauend Schritte der Angriffsdetektion und der Wirkungsminderung zur Wiederherstellung grundlegender P2P-Funktionalität. In Abhängigkeit vom betrachteten P2P-Modell schlagen wir Gegenmaßnahmen vor, die (a) wirksam gegen eine bestimmte Angriffsart einschließlich möglicher Varianten sind, (b) leichtgewichtig in der Ausführung sind, (c) vollständig dezentralisiert, also unabhängig von zentralen Entitäten, sind und (d) reaktive und proaktive Ansätze der Wirkungsminderung ermöglichen. Unsere theoretische Analyse und Simulationen zeigen, dass die vorgestellten Mechanismen zur Detektion und Wirkungsminderung eine Detektionsgenauigkeit von 90-100% erzielen können und dabei selbst in massiven Angriffsszenarien einen niedrigen Mehraufwand von 5-10% erfordern.German
Export:
Suche nach Titel in: TUfind oder in Google

Optionen (nur für Redakteure)

View Item View Item