Online advertising is at the core of today’s Web: it is the main business model, generating large annual revenues expressed in tens of billions of dollars that sponsor most of the online content and services. Online advertising consists of delivering marketing messages, embedded into Web content, to a targeted audience. In this model, entities attract Web traffic by offering the content and services for free and charge advertisers for including advertisements in this traffic (i.e., advertisers pay for users’ attention and interests). Online advertising is a very successful form of advertising as it allows for advertisements (ads) to be targeted to individual users’ interests; especially when advertisements are served on users’ mobile devices, as ads can be targeted to users’ locations and the corresponding context. However, online advertising also introduces a number of problems. Given the high ad revenue at stake, fraudsters have economic incentives to exploit the ad system and generate profit from it. Unfortunately, to achieve this goal, they often compromise users’ online security (e.g., via malware, phishing, etc.). For the purpose of maximizing the revenue by matching ads to users’ interests, a number of techniques are deployed, aimed at tracking and profiling users’ digital footprints, i.e., their behavior in the digital world. These techniques introduce new threats to users’ privacy. Consequently, some users adopt ad-avoidance tools that prevent the download of advertisements and partially thwart user profiling. Such user behavior, as well as exploits of ad systems, have economic implications as they undermine the online advertising business model. Meddling with advertising revenue disrupts the current economic model of the Web, the consequences of which are unclear. Given that today’s Web model relies on online advertising revenue in order for users to have access and consume content and services for “free”, coupled with the fact that there are many threats that could jeopardize this model, in this thesis we address the security, privacy and economic issues stemming from this fundamental element of the Web. In the first part of the thesis, we investigate the vulnerabilities of online advertising systems. We identify how an adversary can exploit the ad system to generate profit for itself, notably by performing inflight modification of ad traffic. We provide a proof-of-concept implementation of the identified threat on Wi-Fi routers. We propose a collaborative approach for securing online advertising and Web browsing against such threats. By investigating how a certificate-based authentication is deployed in practice, we assess the potential of relying on certificate-based authentication as a building block of a solution to protect the ad revenue. We propose a multidisciplinary approach for improving the current state of certificate-based authentication on the Web. In the second part of the thesis, we study the economics of ad systems’ exploits and certain potential countermeasures. We evaluate the potential of different solutions aimed at protecting ad revenue being implemented by the stakeholders (e.g., Internet Service Providers or ad networks) and the conditions under which this is likely to happen. We also study the economic ramifications of ad-avoidance technologies on the monetization of online content. We use game-theory to model the strategic behavior of involved entities and their interactions. In the third part of the thesis, we focus on privacy implications of online advertising. We identify a novel threat to users’ location privacy that enables service providers to geolocate users with high accuracy, which is needed to serve location-targeted ads for local businesses. We draw attention to the large scale of the threat and the potential impact on users’ location privacy.
EPFL2013
