No abstract

Abstract. Existing work on data collection and analysis for aggregation is mainly focused on confidentiality issues. That is, the untrusted Aggregator learns only the aggregation result without divulging individual data inputs. In this paper we extend the existing models with stronger security requirements. Apart from the privacy requirements with respect to the individual inputs, we ask for unforgeability for the aggregate result. We first define the new security requirements of the model. We also instantiate a protocol for private and unforgeable aggregation for multiple independent users. I.e, multiple unsynchronized users owing to personal sensitive information without interacting with each other, contribute their values in a secure way: The Aggregator learns the result of a function without learning individual values, and moreover, it constructs a proof that is forwarded to a verifier that will convince the latter for the correctness of the computation. Our protocol is provably secure in the random oracle model.

Abstract. With the advent of networking applications collecting user data on a massive scale, the privacy of individual users appears to be a major concern. The main challenge is the design of a solution that allows the data analyzer to compute global statistics over the set of individual inputs that are protected by some confidentiality mechanism. Joye et al. [8] recently suggested a solution that allows a centralized party to compute the sum of encrypted inputs collected through a smart metering network. The main shortcomings of this solution are its reliance on a trusted dealer for key distribution and the need for frequent key updates. In this paper we introduce a secure protocol for aggregation of timeseries data that is based on the Joye et al.[8] scheme and in which the main shortcomings of the latter, namely, the requirement for key updates and for the trusted dealer are eliminated. Moreover our scheme supports a dynamic group management, whereby as opposed to Joye et al.[8] leave and join operations do not trigger a key update at the users.

Abstract-Current applications tend to use personal sensitive information to achieve better quality with respect to their services. Since the third parties are not trusted the data must be protected such that individual data privacy is not compromised but at the same time operations on it would be compatible. A wide range of data analysis operations entails a similarity detection algorithm between user data. For instance clustering on big data groups together objects based on the heuristic that similar objects are likely to be put under the same cluster. Similarity decisions are important for numerous applications such as: online social networks, recommendations systems and behavioral advertisement. In this paper we propose a mechanism that protects user privacy and preserves data similarity results although encrypted. We analyze the security of the scheme and we further demonstrate its correctness and feasibility through a real life experiment where "personality traits" by users are collected for a 4square application.

We seek to answer the following question: To what extent can we deduplicate replicated storage ? To answer this question, we design ReDup, a secure storage system that provides users with strong integrity, reliability, and transparency guarantees about data that is outsourced at cloud storage providers. Users store multiple replicas of their data at different storage servers, and the data at each storage server is deduplicated across users. Remote data integrity mechanisms are used to check the integrity of replicas. We consider a strong adversarial model, in which collusions are allowed between storage servers and also between storage servers and dishonest users of the system. A cloud storage provider (CSP) could store less replicas than agreed upon by contract, unbeknownst to honest users. ReDup defends against such adversaries by making replica generation to be time consuming so that a dishonest CSP cannot generate replicas on the fly when challenged by the users. In addition, ReDup employs transparent deduplication, which means that users get a proof attesting the deduplication level used for their files at each replica server, and thus are able to benefit from the storage savings provided by deduplication. The proof is obtained by aggregating individual proofs from replica servers, and has a constant size regardless of the number of replica servers. Our solution scales better than state of the art and is provably secure under standard assumptions.