In the last decade, mobile devices have spread rapidly, becoming more and more part of our everyday lives; this is due to their feature-richness, mobility, and affordable price. At the time of writing, Android is the leader of the market among operating systems, with a share of 76% and two and a half billion active Android devices around the world. Given that such small devices contain a massive amount of our private and sensitive information, the economic interests in the mobile ecosystem skyrocketed. For this reason, not only legitimate apps running on mobile environments have increased dramatically, but also malicious apps have also been on a steady rise. On the one hand, developers of mobile operating systems learned from security mistakes of the past, and they made significant strides in blocking those threats right from the start. On the other hand, these high-security levels did not deter attackers. In this thesis, I present my research contribution about the most meaningful attack and defense scenarios in the userland of the modern Android operating system. I have emphasized "userland'' because attack and defense solutions presented in this thesis are executing in the userspace of the operating system, due to the fact that Android is slightly different from traditional operating systems. After the necessary technical background, I show my solution, RmPerm, in order to enable Android users to better protect their privacy by selectively removing permissions from any app on any Android version. This operation does not require any modification to the underlying operating system because we repack the original application. Then, using again repackaging, I have developed Obfuscapk; it is a black-box obfuscation tool that can work with every Android app and offers a free solution with advanced state of the art obfuscation techniques -- especially the ones used by malware authors. Subsequently, I present a machine learning-based technique that focuses on the identification of malware in resource-constrained devices such as Android smartphones. This technique has a very low resource footprint and does not rely on resources outside the protected device. Afterward, I show how it is possible to mount a phishing attack -- the historically preferred attack vector -- by exploiting two recent Android features, initially introduced in the name of convenience. Although a technical solution to this problem certainly exists, it is not solvable from a single entity, and there is the need for a push from the entire community. But sometimes, even though there exists a solution to a well-known vulnerability, developers do not take proper precautions. In the end, I discuss the Frame Confusion vulnerability; it is often present in hybrid apps, and it was discovered some years ago, but I show how it is still widespread. I proposed a methodology, implemented in the FCDroid tool, for systematically detecting the Frame Confusion vulnerability in hybrid Android apps. The results of an extensive analysis carried out through FCDroid on a set of the most downloaded apps from the Google Play Store prove that 6.63% (i.e., 1637/24675) of hybrid apps are potentially vulnerable to Frame Confusion. The impact of such results on the Android users' community is estimated in 250.000.000 installations of vulnerable apps.

Novel Attacks and Defenses in the Userland of Android

AONZO, Simone
2020-01-15

Abstract

In the last decade, mobile devices have spread rapidly, becoming more and more part of our everyday lives; this is due to their feature-richness, mobility, and affordable price. At the time of writing, Android is the leader of the market among operating systems, with a share of 76% and two and a half billion active Android devices around the world. Given that such small devices contain a massive amount of our private and sensitive information, the economic interests in the mobile ecosystem skyrocketed. For this reason, not only legitimate apps running on mobile environments have increased dramatically, but also malicious apps have also been on a steady rise. On the one hand, developers of mobile operating systems learned from security mistakes of the past, and they made significant strides in blocking those threats right from the start. On the other hand, these high-security levels did not deter attackers. In this thesis, I present my research contribution about the most meaningful attack and defense scenarios in the userland of the modern Android operating system. I have emphasized "userland'' because attack and defense solutions presented in this thesis are executing in the userspace of the operating system, due to the fact that Android is slightly different from traditional operating systems. After the necessary technical background, I show my solution, RmPerm, in order to enable Android users to better protect their privacy by selectively removing permissions from any app on any Android version. This operation does not require any modification to the underlying operating system because we repack the original application. Then, using again repackaging, I have developed Obfuscapk; it is a black-box obfuscation tool that can work with every Android app and offers a free solution with advanced state of the art obfuscation techniques -- especially the ones used by malware authors. Subsequently, I present a machine learning-based technique that focuses on the identification of malware in resource-constrained devices such as Android smartphones. This technique has a very low resource footprint and does not rely on resources outside the protected device. Afterward, I show how it is possible to mount a phishing attack -- the historically preferred attack vector -- by exploiting two recent Android features, initially introduced in the name of convenience. Although a technical solution to this problem certainly exists, it is not solvable from a single entity, and there is the need for a push from the entire community. But sometimes, even though there exists a solution to a well-known vulnerability, developers do not take proper precautions. In the end, I discuss the Frame Confusion vulnerability; it is often present in hybrid apps, and it was discovered some years ago, but I show how it is still widespread. I proposed a methodology, implemented in the FCDroid tool, for systematically detecting the Frame Confusion vulnerability in hybrid Android apps. The results of an extensive analysis carried out through FCDroid on a set of the most downloaded apps from the Google Play Store prove that 6.63% (i.e., 1637/24675) of hybrid apps are potentially vulnerable to Frame Confusion. The impact of such results on the Android users' community is estimated in 250.000.000 installations of vulnerable apps.
cybersecurity; android;
File in questo prodotto:
File Dimensione Formato  
phdunige_3338799.pdf

accesso aperto

Descrizione: Simone Aonzo PhD Thesis
Tipologia: Tesi di dottorato
Dimensione 3.77 MB
Formato Adobe PDF
3.77 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11567/990743
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact