Onwuzurike, Lucky; (2019) Measuring and Mitigating Security and Privacy Issues on Android Applications. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text thesis.pdf - Submitted Version Download (1MB) | Preview

Abstract

Over time, the increasing popularity of the Android operating system (OS) has resulted in its user-base surging past 1 billion unique devices. As a result, cybercriminals and other non-criminal actors are attracted to the OS due to the amount of user information they can access. Aiming to investigate security and privacy issues on the Android ecosystem, previous work has shown that it is possible for malevolent actors to steal users' sensitive personal information over the network, via malicious applications, or vulnerability exploits etc., presenting proof of concepts or evidences of exploits. Due to the ever-changing nature of the Android ecosystem and the arms race involved in detecting and mitigating malicious applications, it is important to continuously examine the ecosystem for security and privacy issues. This thesis presents research contributions in this space, and it is divided into two parts. The first part focuses on measuring and mitigating vulnerabilities in applications due to poor implementation of security and privacy protocols. In particular, we investigate the implementation of the SSL/TLS protocol validation logic, and properties such as ephemerality, anonymity, and end-to-end encryption. We show that, despite increased awareness of vulnerabilities in SSL/TLS implementation by application developers, these vulnerabilities are still present in popular applications, allowing malicious actors steal users' information. To help developers mitigate them, we provide useful recommendations such as enabling SSL/TLS pinning and using the same certificate validation logic in their test and development environments. The second part of this thesis focuses on the detection of malicious applications that compromise users' security and privacy, the detection performance of the different program analysis approach, and the influence of different input generators during dynamic analysis on detection performance. We present a novel method for detecting malicious applications, which is less susceptible to the evolution of the Android ecosystem (i.e., changes in the Android framework as a result of the addition/removal of API calls in new releases) and malware (i.e., changes in techniques to evade detection) compared to previous methods. Overall, this thesis contributes to knowledge around Android apps with respect to, vulnerability discovery that leads to loss of users' security and privacy, and the design of robust Android malware detection tools. It highlights the need for continual evaluation of apps as the ecosystem changes to detect and prevent vulnerabilities and malware that results in a compromise of users' security and privacy.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item