Privad: Practical Privacy in Online Advertising

Online advertising is a major economic force in the Internet today, funding a wide variety of websites and services. Today’s deployments, however, erode privacy and degrade performance as browsers wait for ad networks to deliver ads. This paper presents Privad, an online advertising system designed to be faster and more private than existing systems while filling the practical market needs of targeted advertising: ads shown in web pages; targeting based on keywords, demographics, and interests; ranking based on auctions; view and click accounting; and defense against click-fraud. Privad occupies a point in the design space that strikes a balance between privacy and practical considerations. This paper presents the design of Privad, and analyzes the pros and cons of various design decisions. It provides an informal analysis of the privacy properties of Privad. Based on microbenchmarks and traces from a production advertising platform, it shows that Privad scales to present-day needs while simultaneously improving users’ browsing experience and lowering infrastructure costs for the ad network. Finally, it reports on our implementation of Privad and deployment of over two thousand clients.