Efficient privacy-preserving configurationless service discovery supporting multi-link networks

Data is the pollution problem of the information age, and protecting privacy is the environmental challenge — this quotation from Bruce Schneier laconically illustrates the importance of protecting privacy. Protecting privacy — as well as protecting our planet — is fundamental for humankind. Privacy is a basic human right, stated in the 12th article of the United Nations’ Universal Declaration of Human Rights. The necessity to protect human rights is unquestionable. Nothing ever threatened privacy on a scale comparable to today’s interconnected computers. Ranging from small sensors over smart phones and notebooks to large compute clusters, they collect, generate and evaluate vast amounts of data. Often, this data is distributed via the network, not only rendering it accessible to addressees, but also — if not properly secured — to malevolent parties. Like a toxic gas, this data billows through networks and suffocates privacy. This thesis takes on the challenge of protecting privacy in the area of configurationless service discovery. Configurationless service discovery is a basis for user-friendly applications. It brings great benefits, allowing the configurationless network setup for various kinds of applications; e.g. for communicating, sharing documents and collaborating, or using infrastructure devices like printers. However, while today’s various protocols provide some means of privacy protection, typical configurationless service discovery solutions do not even consider privacy. As configurationless service discovery solutions are ubiquitous and run on almost every smart device, their privacy problems affect almost everyone. The quotation aligns very well with configurationless service discovery. Typically, configurationless service discovery solutions realize configurationlessness by using cleartext multicast messages literally polluting the local network and suffocating privacy. Messages containing private cleartext data are sent to everyone, even if they are only relevant for a few users. The typical means for mitigating the network pollution problem caused by multicast per se, regardless of the privacy aspects, is confining multicast messages to a single network link or to the access network of a WiFi access point; institutions often even completely deactivate multicast. While this mitigates the privacy problem, it also strongly scales configurationless service discovery down, either confining it or rendering it completely unusable. In this thesis, we provide an efficient configurationless service discovery framework that protects the users’ privacy. It further reduces the network pollution by reducing the number of necessary multicast messages and offers a mode of operation that is completely independent of multicast. Introducing a multicast independent mode of operation, we also address the problem of the limited range in which services are discoverable. Our framework comprises components for device pairing, privacy-preserving service discovery, and multi-link scaling. These components are independent and — while usable in a completely separated way — are meant to be used as an integrated framework as they work seamlessly together. Based on our device pairing and privacy-preserving service discovery components, we published IETF Internet drafts specifying a privacy extension for DNS service discovery over multicast DNS, a wildly used protocol stack for configurationless service discovery. As our drafts have already been adopted by the dnssd working group, they are likely to become standards