University of Crete (UOC)
Online social networks from a malicious perspective: novel attack techniques and defense mechanisms
Abstract
dc:descriptionSocial networking services have become the most popular digital services, occupying themajority of the time users spend online. These services have greatly evolved from the firstgeneration of social networks, and offer an expansive set of functionality ranging from userinteraction and content sharing, to online gaming and single sign-on services. These serviceshave inadvertently and irrevocably affected the World Wide Web, and forever altered thenotion of privacy in the digital era. A natural consequence of their popularity was to alsodraw the attention of the Internet miscreants that target users for profit.The vast amounts of personal information and interests that users divulge in these services,along with the high amount of trust users implicitly show to communication received withinsuch networks, has rendered online social networks the ideal springboard for deploying highlyprofitablepersonalized attacks. Attacks in social networks can build upon the expertise ofmore traditional attack vectors (e.g., email spam) and also incorporate novel techniques forcreating complex and intricate attacks. The ever-evolving nature of these networks and thecontinuous incorporation of novel functionality introduces new design vulnerabilities whichcan be exploited by adversaries.Security research in social networks mandates that researchers assume the role of theadversary when exploring the security aspects of these services. Their proprietary naturerestricts their deployment in the controlled environment of a laboratory, and may requirea black-box testing approach as their internal mechanisms are often unknown. As such,researchers must interact with the actual services and their users. Only then will they be ableto “anticipate” techniques that adversaries may employ in the future, and develop effectivedefense mechanisms that will hinder the actual attacks.The dissertation demonstrates that by misusing functionality found in various onlineservices and social networks, one can build and deploy effective novel attacks. The resultsof the practical experiments reveal the vulnerable design of existing defense mechanismsemployed by social networks and their inability to protect their assets from adversaries. Thecharacteristics of the attack techniques and the outcome of the experiments guide the designand implementation of new defense mechanisms.Specifically, we identify the following resources as the “assets” of social networking services,which should be protected against adversaries: (i) user information, (ii) user accounts and(iii) user actions. We assume the role of the attacker and deploy attacks that bypass anymechanisms (if any) designed to protect each type of asset. First, we explore various techniquesfor harvesting and correlating (personal) user information that can be used for craftingpersonalized attacks. Next, we demonstrate the effectiveness of automated attacks againphoto-based authentication mechanisms designed to hinder adversaries from compromisinguser accounts. Finally, we conduct extensive experiments to explore the defense mechanismsdeployed by social networks to detect and remove actions by malicious users in regards tolocation-based functionality. In each case, based on the insight gained from the experimentswe design mechanisms for mitigating or (if possible) preventing these novel attacks.
Degree
thesis:*- Grantor dc:publisher
- University of Crete (UOC)
- Year dc:date
- 2014
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Polakis, Iasonas-Stylianos
- Πολάκης, Ιάσων-Στυλιανός
Subjects
dc:subject × 20- Ιστοσελίδες κοινωνικής δικτύωσης
- Ασφάλεια
- Τεχνικές επιθέσεων
- Αμυντικοί μηχανισμοί
- Λειτουργικότητα βάση γεωγραφικής τοποθεσίας του χρήστη
- Μηχανισμοί αυθεντικοποίησης
- Ιδιωτικότητα χρηστών
- Συλλογή δεδομένων
- Online social networks
- Security
- Attack techniques
- Defense mechanisms
- Location-based functionality
- Authentication mechanism
- User privacy
- Data mining
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Computer and Information Sciences
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/36570