Philippe Golle

Controlling data in the cloud: outsourcing computation without outsourcing control

Cloud computing is clearly one of todays most enticing technology areas due, at least in part, to its cost-efficiency and flexibility. However, despite the surge in activity and interest, there are significant, persistent concerns about cloud computing that are impeding momentum and will eventually compromise the vision of cloud computing as a new IT procurement model. In this paper, we characterize the problems and their impact on adoption. In addition, and equally importantly, we describe how the combination of existing research thrusts has the potential to alleviate many of the concerns impeding adoption. In particular, we argue that with continued research advances in trusted computing and computation-supporting encryption, life in the cloud can be advantageous from a business intelligence standpoint over the isolated alternative that is more common today.

Incentives for sharing in peer-to-peer networks

We consider the free-rider problem that arises in peer-to-peer file sharing networks such as Napster: the problem that individual users are provided with no incentive for adding value to the network. We examine the design implications of the assumption that users will selfishly act to maximize their own rewards, by constructing a formal game theoretic model of the system and analyzing equilibria of user strategies under several novel payment mechanisms. We support and extend upon our theoretical predictions with experimental results from a multi-agent reinforcement learning model.

Secure Conjunctive Keyword Search over Encrypted Data

We study the setting in which a user stores encrypted documents (e.g. e-mails) on an untrusted server. In order to retrieve documents satisfying a certain search criterion, the user gives the server a capability that allows the server to identify exactly those documents. Work in this area has largely focused on search criteria consisting of a single keyword. If the user is actually interested in documents containing each of several keywords (conjunctive keyword search) the user must either give the server capabilities for each of the keywords individually and rely on an intersection calculation (by either the server or the user) to determine the correct set of documents, or alternatively, the user may store additional information on the server to facilitate such searches. Neither solution is desirable; the former enables the server to learn which documents match each individual keyword of the conjunctive search and the latter results in exponential storage if the user allows for searches on every set of keywords.

Universal Re-encryption for Mixnets

We introduce a new cryptographic technique that we call universal re-encryption. A conventional cryptosystem that permits re-encryption, such as ElGamal, does so only for a player with knowledge of the public key corresponding to a given ciphertext. In contrast, universal re-encryption can be done without knowledge of public keys. We propose an asymmetric cryptosystem with universal re-encryption that is half as efficient as standard ElGamal in terms of computation and storage.

On the Anonymity of Home/Work Location Pairs

Many applications benefit from user location data, but location data raises privacy concerns. Anonymization can protect privacy, but identities can sometimes be inferred from supposedly anonymous data. This paper studies a new attack on the anonymity of location data. We show that if the approximate locations of an individuals home and workplace can both be deduced from a location trace, then the median size of the individuals anonymity set in the U.S. working population is 1, 21 and 34,980, for locations known at the granularity of a census block, census track and county respectively. The location data of people who live and work in different regions can be re-identified even more easily. Our results show that the threat of re-identification for location data is much greater when the individuals home and work locations can both be deduced from the data. To preserve anonymity, we offer guidance for obfuscating location traces before they are disclosed.

Uncheatable Distributed Computations

Computationally expensive tasks that can be parallelized are most efficiently completed by distributing the computation among a large number of processors. The growth of the Internet has made it possible to invite the participation of just about any computer in such distributed computations. This introduces the potential for cheating by untrusted participants. In a commercial setting where participants get paid for their contribution, there is incentive for dishonest participants to claim credit for work they did not do. In this paper, we propose security schemes that defend against this threat with very little overhead. Our weaker scheme discourages cheating by ensuring that it does not pay off, while our stronger schemes let participants prove that they have done most of the work they were assigned with high probability.

Machine learning attacks against the Asirra CAPTCHA

The Asirra CAPTCHA [EDHS2007], proposed at ACM CCS 2007, relies on the problem of distinguishing images of cats and dogs (a task that humans are very good at). The security of Asirra is based on the presumed difficulty of classifying these images automatically. In this paper, we describe a classifier which is 82.7% accurate in telling apart the images of cats and dogs used in Asirra. This classifier is a combination of support-vector machine classifiers trained on color and texture features extracted from images. Our classifier allows us to solve a 12-image Asirra challenge automatically with probability 10.3%. This probability of success is significantly higher than the estimate of 0.2% given in [EDHS2007] for machine vision attacks. Our results suggest caution against deploying Asirra without safeguards. We also investigate the impact of our attacks on the partial credit and token bucket algorithms proposed in [EDHS2007]. The partial credit algorithm weakens Asirra considerably and we recommend against its use. The token bucket algorithm helps mitigate the impact of our attacks and allows Asirra to be deployed in a way that maintains an appealing balance between usability and security. One contribution of our work is to inform the choice of safeguard parameters in Asirra deployments.

Revisiting the uniqueness of simple demographics in the US population

According to a famous study [10] of the 1990 census data, 87% of the US population can be uniquely identified by gender, ZIP code and full date of birth. This short paper revisits the uniqueness of simple demographics in the US population based on the most recent census data (the 2000 census). We offer a detailed, comprehensive and up-to-date picture of the threat to privacy posed by the disclosure of simple demographic information. Our results generally agree with the findings of [10], although we find that disclosing ones gender, ZIP code and full date of birth allows for unique identification of fewer individuals (63% of the US population) than reported in [10]. We hope that our study will be a useful reference for privacy researchers who need simple estimates of the comparative threat of disclosing various demographic data.

Dining Cryptographers Revisited

Dining cryptographers networks (or DC-nets) are a privacy-preserving primitive devised by Chaum for anonymous message publication. A very attractive feature of the basic DC-net is its non-interactivity. Subsequent to key establishment, players may publish their messages in a single broadcast round, with no player-to-player communication. This feature is not possible in other privacy-preserving tools like mixnets. A drawback to DC-nets, however, is that malicious players can easily jam them, i.e., corrupt or block the transmission of messages from honest parties, and may do so without being traced.

Incentives for Sharing in Peer-to-Peer Networks

We consider the free-rider problem in peer-to-peer file sharing networks such as Napster: that individual users are provided with no incentive for adding value to the network. We examine the design implications of the assumption that users will selfishly act to maximize their own rewards, by constructing a formal game theoretic model of the system and analyzing equilibria of user strategies under several novel payment mechanisms. We support and extend this workwith results from experiments with a multi-agent reinforcement learning model.