Flocks: Distributed Proxies for Browsing Privacy

Olivier

2004

Citation information

Abstract

This paper introduces a Privacy-Enhancing Technology (PET) based on a hybrid of Crowds and anonymising proxies. The PET — referred to as Flocks — operates by establishing a number of Web proxies and letting these proxies randomly forward requests to other proxies (or the final destination). This distributes users’ requests over a number of such proxies, thereby helping to protect their (browsing) privacy.

The problem that the paper considers is the effect of two primary design parameters on the privacy of the overall system. These parameters are the probability with which a proxy sends a request to the destination server rather than another proxy (α) and the number of proxies in the system (N). Two privacy objectives are identified, namely the number of hops used to satisfy a request and the portion of proxies that ‘know’ about a request. A third requirement deals with the external communication costs of the system. A formal analysis is performed to determine these three factors from the two identified parameters. Finally numerical examples are used to explore the impact of these two parameter choices in concrete terms.

The proposed PET differs from existing PETs in two significant manners: It is primarily intended to be used inside an organisation to protect the privacy of users inside the organisation (in particular, employees) and it takes explicit cognisance of forensic factors.

Full text

Definitive version

BibTeX reference