We study the cryptographic complexity of two-party differentially-private protocols for a large natural class of boolean functionalities. Information theoretically, McGregor et al. [FOCS 2010] and Goyal et al. [Crypto 2013] demonstrated several functionalities for which the maximal possible accuracy in the distributed setting is significantly lower than that in the client-server setting. Goyal et al. [Crypto 2013] further showed that "highly accurate" protocols in the distributed setting for any non-trivial functionality in fact imply the existence of one-way functions. However, it has remained an open problem to characterize the exact cryptographic complexity of this class. In particular, we know that semi-honest oblivious transfer helps obtain optimally accurate distributed differential privacy. But we do not know whether the reverse is true. We study the following question: Does the existence of optimally accurate distributed differentially private protocols for any class of functionalities imply the existence of oblivious transfer (or equivalently secure multi-party computation)? We resolve this question in the affirmative for the class of boolean functionalities that contain an XOR embedded on adjacent inputs. We give a reduction from oblivious transfer to: - Any distributed optimally accurate epsilon-differentially private protocol with epsilon > 0 computing a functionality with a boolean XOR embedded on adjacent inputs. - Any distributed non-optimally accurate epsilon-differentially private protocol with epsilon > 0, for a constant range of non-optimal accuracies and constant range of values of epsilon, computing a functionality with a boolean XOR embedded on adjacent inputs. Enroute to proving these results, we demonstrate a connection between optimally-accurate twoparty differentially-private protocols for functions with a boolean XOR embedded on adjacent inputs, and noisy channels, which were shown by Crépeau and Kilian [FOCS 1988] to be sufficient for oblivious transfer.
We study the cryptographic complexity of two-party differentially-private protocols for a large natural class of boolean functionalities. Information theoretically, McGregor et al. [FOCS 2010] and Goyal et al. [Crypto 2013] demonstrated several functionalities for which the maximal possible accuracy in the distributed setting is significantly lower than that in the client-server setting. Goyal et al. [Crypto 2013] further showed that "highly accurate" protocols in the distributed setting for any non-trivial functionality in fact imply the existence of one-way functions. However, it has remained an open problem to characterize the exact cryptographic complexity of this class. In particular, we know that semi-honest oblivious transfer helps obtain optimally accurate distributed differential privacy. But we do not know whether the reverse is true. We study the following question: Does the existence of optimally accurate distributed differentially private protocols for any class of functionalities imply the existence of oblivious transfer (or equivalently secure multi-party computation)? We resolve this question in the affirmative for the class of boolean functionalities that contain an XOR embedded on adjacent inputs. We give a reduction from oblivious transfer to: - Any distributed optimally accurate epsilon-differentially private protocol with epsilon > 0 computing a functionality with a boolean XOR embedded on adjacent inputs. - Any distributed non-optimally accurate epsilon-differentially private protocol with epsilon > 0, for a constant range of non-optimal accuracies and constant range of values of epsilon, computing a functionality with a boolean XOR embedded on adjacent inputs. Enroute to proving these results, we demonstrate a connection between optimally-accurate twoparty differentially-private protocols for functions with a boolean XOR embedded on adjacent inputs, and noisy channels, which were shown by Crépeau and Kilian [FOCS 1988] to be sufficient for oblivious transfer.
We study the maximal achievable accuracy of distributed differentially private protocols for a large natural class of boolean functions, in the computational setting. In the information theoretic model, McGregor et al. [FOCS 2010] and Goyal et al. [CRYPTO 2013] have demonstrated several functionalities whose differentially private computation results in much lower accuracies in the distributed setting, as compared to the client-server setting. We explore lower bounds on the computational assumptions under which this particular accuracy gap can possibly be reduced for general two-party boolean output functions. In the distributed setting, it is possible to achieve optimal accuracy, i.e. the maximal achievable accu-racy in the client-server setting, for any function, if a semi-honest secure protocol for oblivious transfer exists. However, we show the following strong impossibility results: ◦ For any boolean function and fixed level of privacy, the maximal achievable accuracy of any (fully) black-box construction based on existence of key-agreement protocols is at least a constant smaller than optimal achievable accuracy. Since key-agreement protocols imply the existence of one-way functions, this separation also extends to one-way functions.
Which computational complexity assumptions are inherent to cryptography? We present a broad framework to pose and investigate this question. We first aim to understand the “cryptographic complexity ” of various tasks, independent of any computational assumptions. In our framework the cryptographic tasks are modeled as multi-party computation functionalities. We consider a universally composable secure protocol for one task given access to another task as the most natural complexity reduction between the two tasks. Some of these cryptographic complexity reductions are unconditional, others are unconditionally non-existent, but the vast majority appear to depend on computational assumptions; it is this relationship with computational assumptions that we study. In our detailed investigation of large classes of 2-party functionalities, we find that every reduction we are able to classify turns out to be unconditionally true or false, or else equivalent to the existence of one-way functions (OWF) or of semi-honest (equivalently, standalone-secure) oblivious transfer protocols (sh-OT). This leads us to conjecture that there are only a small finite number of distinct (i.e., black-box separable) computational assump-tions that are inherent among the infinite number of different cryptographic reductions in our framework. On moving to 3-party functionalities, there does exist a natural computational assumption, distinct from the OWF and sh-OT assumptions (namely, the existence of a key agreement protocol) that manifests itself as a cryptographic
In cryptographic protocols, honest parties would prefer that their security is assured even in presence of adversarial parties who have unbounded computational power. Information theoretic secure realization of cryptographic primitives provides such guarantees; but for most tasks such strong security guarantees cannot be provided for any reasonable notion of security. The standard technique used in cryptography is to assume the existence of some puzzle, whose hard instances are easy to generate but no efficient algorithm can solve them. Such assumptions, which define intractable problems for efficient algorithms, are called computational intractability assumptions.
In this work, we motivate a study of computational intractability assumptions which is goal driven and lends support to the fundamental nature of some of the traditional assumptions beyond being historical accidents. Secure multi-party computation deals with the study of constructing secure protocols for general cryptographic tasks which conform to various notions of security. Inspired by complexity theory, we use the notion of reduction to further our understanding of computational intractability assumptions. Our framework explores the hardness of natural assumptions of the form: “Task F can be securely computed given an ideally secure facility for computing G ”.
Formally, we characterize the minimal computational intractability assumption which is sufficient to securely realize a functionality using trusted copies of some other functionality. A functionality F reduces to G , similar to the complexity-theoretic notion of reduction, if F can be securely realized when parties can access trusted copies of G . Thus, if F does not reduce to G information-theoretically, we are interested in characterizing the minimal assumption, which is sufficient, to securely realize the reduction of F to G .
In our framework, we explore the relative strengths of these minimal computational assumptions. The fundamental problem, then, is to establish relations, like implication, equivalence and separation, among the assumptions which correspond to various reductions. In this work, we further our understanding of the minimal computational assumptions corresponding to such reductions by showing the following results:
1. We identify a spectrum for the set of assumptions that correspond to reduction between a pair of tasks under different security notions. These reductions are implied by the computational intractability assumption: “semi-honest secure protocol for oblivious transfer exists” (sh-OT assumption). Additionally, the information theoretically impossible reductions considered in this thesis imply the assumption: “one-way functions exist” (OWF assumption). We conjecture that OWF assumption is necessary for these reductions. In fact, the reductions considered in this thesis are either information theoretically true, false, equivalent to OWF assumption or equivalent to sh-OT assumption [MPR10a].
We expand our study to encompass reductions involving randomized functionalities and study the consequences of providing parties with a trusted source of common unbiased coins. For a strong notion of security they turn out to be useless and unless sh-OT assumption holds these randomness sources are not useful [MOPR11, MP11].
2. We [MPS10] seek to lend additional support to the widely believed premise that “non-trivial cryptography entails the existence of one-way functions”. We show that (constant round) weak coin-tossing protocols, imply OWF assumption. For the general problem, i.e. secure weak coin-tossing protocols with polynomial round complexity, we show a slightly weaker implication NP n BPP.
3. Finally, we indicate the possibility of a wide range of intractability assumptions in our spectrum, intermediate to OWF assumption and sh-OT assumption. We show that a class of natural assumptions corresponding to reductions that are relativistically separated from OWF assumption. Similar techniques have been used to show their separation from the assumption that “public-key encryption exists” (PKE assumption) [MMP11]; but these assumptions are not known to imply sh-OT assumption. We conjecture that, in fact, sh-OT assumption is relativistically separated from these assumptions.
In the setting of secure two-party computation, two mutually distrusting parties wish to compute some function of their inputs while preserving, to the extent possible, various security properties such as privacy, correctness, and more. One desirable property is fairness which guarantees, informally, that if one party receives its output, then the other party does too. Cleve (STOC 1986) showed that complete fairness cannot be achieved in general without an honest majority. Since then, the accepted folklore has been that nothing non-trivial can be computed with complete fairness in the two-party setting, and the problem has been treated as closed since the late ’80s. In this paper, we demonstrate that this folklore belief is false by showing completely-fair protocols for various non-trivial functions in the two-party setting based on standard cryptographic assumptions. We first show feasibility of obtaining complete fairness when computing any function over polynomial-size domains that does not contain an “embedded XOR”; this class of functions includes boolean AND/OR as well as Yao’s “millionaires ’ problem”. We also demonstrate feasibility for certain functions that do contain an embedded XOR, and prove a lower bound showing that any completely-fair protocol for such functions must have round complexity super-logarithmic in the security parameter. Our results demonstrate that the question of completely-fair secure computation without an honest majority is far from closed.
In the setting of secure two-party computation, two mutually distrusting parties wish to compute some function of their inputs while preserving, to the extent possible, various security properties such as privacy, correctness, and more. One desirable property is fairness which guarantees, informally, that if one party receives its output, then the other party does too. Cleve [1986] showed that complete fairness cannot be achieved in general without an honest majority. Since then, the accepted folklore has been that nothing non-trivial can be computed with complete fairness in the two-party setting. We demonstrate that this folklore belief is false by showing completely fair protocols for various nontrivial functions in the two-party setting based on standard cryptographic assumptions. We first show feasibility of obtaining complete fairness when computing any function over polynomial-size domains that does not contain an “embedded XOR”; this class of functions includes boolean AND/OR as well as Yao’s “millionaires’ problem”. We also demonstrate feasibility for certain functions that do contain an embedded XOR, though we prove a lower bound showing that any completely fair protocol for such functions must have round complexity super-logarithmic in the security parameter. Our results demonstrate that the question of completely fair secure computation without an honest majority is far from closed.
Secure two-party computation allows two parties to evaluate a function on their private inputs while keeping all information private except what can be inferred from the outputs. A major building block and the foundation for many efficient secure computation protocols is oblivious transfer (OT). In an OT protocol a sender inputs two messages (x_{0}, x_{1}) while a receiver with choice bit c wants to receive message x_{c}.The OT protocol execution guarantees that the sender learns no information about c and the receiver learns no information about x_{1−c}. This thesis focuses on the efficient generation of OTs and their use in secure computation. In particular, we show how to compute OTs more efficiently, improve generic secure computation protocols which can be used to securely evaluate any functionality, and develop highly efficient special-purpose protocols for private set intersection (PSI). We outline our contributions in more detail next. More Efficient OT Extensions. The most efficient OT protocols are based on public-key cryptography and require a constant number of exponentiations per OT. However, for many practical applications where millions to billions of OTs need to be computed, these exponentiations become prohibitively slow. To enable these applications, OT extension protocols [Bea96, IKNP03] can be used, which extend a small number of public-key-based OTs to an arbitrarily large number using cheap symmetric-key cryptography only. We improve the computation and communication efficiency of OT extension protocols and show how to achieve security against malicious adversaries, which can arbitrarily deviate from the protocol, at low overhead. Our resulting protocols can compute several million of OTs per second and we show that, in contrast to previous belief, the local computation overhead for computing OTs is so low that the main bottleneck is the network bandwidth. Parts of these results are published in: • G. Asharov, Y. Lindell, T. Schneider, M. Zohner. More Efficient Oblivious Transfer and Extensions for Faster Secure Computation. In 20th ACM Conference on Computer and Communications Security (CCS’13). • G. Asharov, Y. Lindell, T. Schneider, M. Zohner. More Efficient Oblivious Transfer Extensions with Security for Malicious Adversaries. In 34th Advances in Cryptology – EUROCRYPT’15. • G. Asharov, Y. Lindell, T. Schneider, M. Zohner. More Efficient Oblivious Transfer Extensions. To appear in Journal of Cryptology. Online at http://eprint.iacr.org/2016/602. Communication-Efficient Generic Secure Two-Party Computation. Generic secure two-party computation techniques allow to evaluate a function, represented as a circuit of linear (XOR) and non-linear (AND) gates. One of the most prominent generic secure two-party computation protocols is Yao’s garbled circuits [Yao86], for which many optimizations have been proposed. Shortly after Yao’s protocol, the generic secure protocol by Goldreich-Micali-Wigderson (GMW) [GMW87] was introduced. The GMW protocol requires a large number of OTs and was believed to be less efficient for secure two-party computation than Yao’s protocol [HL10, CHK+12]. We improve the efficiency of the GMW protocol and show that it can outperform Yao’s garbled circuits protocol in settings with low bandwidth. Furthermore, we utilize the flexibility of OT and outline special-purpose constructions that can be used within the GMW protocol and which improve its efficiency even further. Parts of these results are published in: • T. Schneider, M. Zohner. GMW vs. Yao? Efficient Secure Two-Party Computation with Low Depth Circuits. In 17th International Conference on Financial Cryptography and Data Security (FC’13). • D. Demmler, T. Schneider, M. Zohner. ABY - A Framework for Efficient Mixed-Protocol Secure Two-Party Computation. In 22th Network and Distributed System Security Symposium (NDSS’15). • G. Dessouky, F. Koushanfar, A.-R. Sadeghi, T. Schneider, S. Zeitouni, M. Zohner. Pushing the Communication Barrier in Secure Computation using Lookup Tables. In 24th Network and Distributed System Security Symposium (NDSS’17). Faster Private Set Intersection (PSI). PSI allows two parties to compute the intersection of their private sets without revealing any element that is not in the intersection. PSI is a well-studied problem in secure computation and many special-purpose protocols have been proposed. However, existing PSI protocols are several orders of magnitude slower than an insecure naive hashing solution that is used in practice. In addition, many protocols were compared in a biased fashion, which makes it difficult to identify the most promising solution for a particular scenario. We systematize the progress made on PSI protocols by reviewing, optimizing, and comparing existing PSI protocols. We then introduce a novel PSI protocol that is based on our efficiency improvements in OT extension protocols and which outperforms existing protocols by up to two orders of magnitude. Parts of these results are published in: • B. Pinkas, T. Schneider, M. Zohner. Faster Private Set Intersection Based on OT Extension. In 23th USENIX Security Symposium (USENIX Security’14). • B. Pinkas, T. Schneider, G. Segev, M. Zohner. Phasing: Private Set Intersection using Permutation-based Hashing. In 24th USENIX Security Symposium (USENIX Security’15). • B. Pinkas, T. Schneider, M. Zohner. Scalable Private Set Intersection Based on OT Extension. Journal paper. In submission. Online at http://iacr.eprint.org/2016/930.