Privacy-preserving mechanisms for targeted mobile advertising

Targeted advertising is a complex mechanism to identify, track and profile users behaviour for their activities over Internet/mobile apps and to serve them advertisements (ads) according to their (profile) interests. Advertising companies develop tracking technologies and embed them into user's Internet browsers and mobile apps, which have raised concerns over privacy issues. Mobile users are uncertain whether to trust targeted ads and advertising companies, however, they tend to view ads relevant to their (profile) interests provided that their information is kept private.This thesis focuses on collection of user's personal information for classification of mobile ads and characterising and quantifying privacy issues arising in mobile ads. In addition, this thesis introduces two novel privacy preserving mechanisms, which address three issues: privacy, control of information, and private user profiling for targeted advertising. First, the ProfileGuard, is an app-based obfuscation mechanism of suggesting obfuscating apps via different strategies, which removes prevalence amongst private interests present in a user profile. Under the assumptions that users will be operating in resource-constrained scenarios, such as fixed 4G data plan, limited processing and battery powers, the ProfileGuard presents, analytically and via experimentations, a good trade-off between the level of efforts required by obfuscating strategies and resulting privacy protection. Second, a framework for privacy preserving mobile advertising system that enables device-based private user profiling, targeted ads and private billing of retrieved/clicked ads. This proposal relies on the use of cryptographic techniques to query ads databases for matching (to profile) ads, without the databases learning the contents or result of queries. This thesis demonstrates, using experimentations, the feasibility of this framework by measuring processing times, communication delays and bandwidth usage, in addition to, its responsiveness and scalability under the scenario where huge number of users operate in parallel. In addition to analytical and experimental evaluations of above mechanisms, this thesis demonstrates the practicality of these mechanisms by developing prototype mobile applications that can be used on smartphones. These mechanisms guarantee to protect user privacy, do not necessitate the re-design of current advertising systems and can operate in parallel to existing systems