We design efficient protocols for processors to extract private randomness over a network with Byzantine faults, when each processor has access to an independent weakly-random n-bit source of sufficient min-entropy.We give several such network extractor protocols in both the information theoretic and computational settings.For a computationally unbounded adversary, we construct protocols in both the synchronous and asynchronous settings.These network extractors imply efficient protocols for leader election (synchronous setting only) and Byzantine agreement which tolerate a linear fraction of faults,even when the min-entropy is only 2(logn)Omega(1).For larger min-entropy,in the synchronous setting the fraction of tolerable faults approaches the bounds in the perfect-randomness case.Our network extractors for a computationally bounded adversary work in the synchronous setting even when 99% of the parties are faulty, assuming trapdoor permutations exist. Further, assuming a strong variant of the Decisional Diffie-Hellman Assumption, we construct a network extractor in which all parties receive private randomness. This yields an efficient protocol for secure multi-party computation with imperfect randomness, when the number of parties is at least polylog (n) and where the parties only have access to an independent source with min-entropy nOmega(1).
We design efficient protocols for processors to extract private randomness over a network with Byzantine faults, when each processor has access to an independent weakly-random n-bit source of sufficient min-entropy.We give several such network extractor protocols in both the information theoretic and computational settings.For a computationally unbounded adversary, we construct protocols in both the synchronous and asynchronous settings.These network extractors imply efficient protocols for leader election (synchronous setting only) and Byzantine agreement which tolerate a linear fraction of faults,even when the min-entropy is only 2(logn)Omega(1).For larger min-entropy,in the synchronous setting the fraction of tolerable faults approaches the bounds in the perfect-randomness case.Our network extractors for a computationally bounded adversary work in the synchronous setting even when 99% of the parties are faulty, assuming trapdoor permutations exist. Further, assuming a strong variant of the Decisional Diffie-Hellman Assumption, we construct a network extractor in which all parties receive private randomness. This yields an efficient protocol for secure multi-party computation with imperfect randomness, when the number of parties is at least polylog (n) and where the parties only have access to an independent source with min-entropy nOmega(1).
We show how to efficiently extract truly random bits from two independent sources of linear min-entropy, under a computational assumption. The assumption we rely on is the existence of an efficiently computable permutation f1, such that for any source X ¿ {0, 1}nwith linear min-entropy, any circuit of size poly(n) cannot invert f(X) with non-negligible probability. Under the stronger assumption that f(X) cannot be inverted even by circuits of size poly(nlog n) with nonnegligible probability, we design a lossless computational network extractor protocol. Namely, we design a protocol for a set of players, each with access to an independent source of linear min-entropy, with the guarantee that at the end of the protocol, each honest player is left with bits that are computationally indistinguishable from being uniform and private. Our protocol succeeds as long as there are at least two honest players. Our results imply that if such one-way permutations exist, and enhanced trapdoor permutations exist, then secure multiparty computation with imperfect randomness is possible for any number of players, as long as at least two of them are honest. We also construct a network extractor protocol for the case where each source has only polynomially-small min-entropy (n¿for some constant ¿ > 0). For this we need at least a constant u(¿) (which depends on ¿) number of honest players, and we need that the one-way permutation is hard to invert even on polynomially small min-entropy sources.
We study the problem of constructing multi-source extractors in the quantum setting, which extract almost uniform random bits against quantum side information collected from several initially independent classical random sources. This is a natural generalization of seeded randomness extraction against quantum side information and classical independent source extraction. With new challenges such as potential entanglement in the side information, it is not a prior clear under what conditions do quantum multi-source extractors exist; the only previous work is [KK12], where the classical inner-product two-source extractors of [CG88] and [DEOR04] are shown to be quantum secure in the restricted Independent Adversary (IA) Model and entangled Bounded Storage (BS) Model. In this paper we propose a new model called General Entangled (GE) Adversary Model, which allows arbitrary entanglement in the side information and subsumes both the IA model and the BS model. We proceed to show how to construct GE-secure quantum multi-source extractors. To that end, we propose another model called One-sided Adversary (OA) Model, which is weaker than all the above models. Somewhat surprisingly, we establish equivalence between strong OA-security and strong GE-security. As a result, all classical multi-source extractors can either directly work, or be modified to work in the GE model at the cost of one extra random source. Thus, our constructions essentially match the best known constructions of classical multi-source extractors. We also apply our techniques to two important problems in cryptography and distributed computing --- privacy amplification and network extractor. We show that as long as the sources have certain amounts of conditional min-entropy in our GE model (even with entangled quantum side information), we can design very efficient privacy amplification protocols and network extractors.
We study the question of basing symmetric key cryptography on weak secrets. In this setting, Alice and Bob share an n-bit secret W, which might not be uniformly random, but the adversary has at least k bits of uncertainty about it (formalized using conditional min-entropy). Since standard symmetrickey primitives require uniformly random secret keys, we would like to construct an authenticated key agreement protocol in which Alice and Bob use W to agree on a nearly uniform key R, by communicating over a public channel controlled by an active adversary Eve. We study this question in the information theoretic setting where the attacker is computationally unbounded. We show that single-round (i.e. one message) protocols do not work when k ≤ n 2, and require poor parameters even when n 2 < k ≪ n. On the other hand, for arbitrary values of k, we design a communication efficient two-round (challenge-response) protocol extracting nearly k random bits. This dramatically improves the previous construction of Renner and Wolf [RW03], which requires Θ(λ + log(n)) rounds where λ is the security parameter. Our solution takes a new approach by studying and constructing “non-malleable” seeded randomness extractors — if an attacker sees a random seed X and comes up with an arbitrarily related seed X ′, then we bound the relationship between R = Ext(W;X) and R ′ = Ext(W;X ′). We also extend our two-round key agreement protocol to the “fuzzy ” setting, where Alice and Bob share “close ” (but not equal) secrets WA and WB, and to the Bounded Retrieval Model (BRM) where the size of the secret W is huge.
Secure multi-party computation (MPC) is a fundamental problem in secure distributed computing. The optimal resilience for perfectly-secure MPC in synchronous and asynchronous networks is t < n/3 and t < n/4 respectively, where n is the number of parties and t is the number of corruptions. A natural question is whether there exists a protocol tolerating ts < n/3 corruptions in a synchronous network and ta < n/4 corruptions in an asynchronous network. We design such a protocol, if 3ts + ta < n. For our protocol, we present a perfectly-secure Byzantine agreement (BA) protocol, tolerating t < n/3 corruptions in any network and a perfectly-secure verifiable secret-sharing (VSS) protocol, tolerating ts and ta corruptions in a synchronous and an asynchronous network respectively.
Abstract. Secure multiparty computation (MPC) allows a set of parties to securely evaluate any agreed function of their inputs, even when up to t of the n parties are faulty. Protocols for synchronous networks (where every sent message is assumed to arrive within a constant time) tolerate up to t < n/2 faulty parties, whereas in the more realistic asynchronous setting (with no a priory information on maximal message delay) only security against t < n/3 is possible. Note that even asynchronous Byzantine agreement requires t < n/3. In this paper, we are interested in the minimal synchronicity assumption for achieving security against t < n/2. It turns out that the bottleneck of asynchronous MPC is the distribution of the inputs: Once the inputs are correctly distributed, any deterministic function can be computed over a fully asynchronous network with t < n/2. Furthermore, we show that the inputs can be verifiably distributed with t < n/2, if a single round of synchronous broadcast is available. Composing the above, we obtain the first MPC protocol that achieves security against t < n/2 without assuming a fully synchronous network. Actually our protocol guarantees security against any faulty minority in an almost asynchronous network, i.e. in a network with one single round of synchronous broadcast (followed by a fully asynchronous communication). Furthermore our protocol takes inputs of all parties (in a fully asynchronous network only inputs of n−t parties can be guaranteed), and so achieves everything that is possible in synchronous networks (but impossible in fully asynchronous networks) at the price of just one synchronous broadcast round. As tools for our protocol we introduce the notions of almost non-interactive verifiable secret-sharing and almost non-interactive zero-knowledge proof of knowledge, which are of independent interest as they can serve as efficient replacements for fully non-interactive verifiable secret-sharing and fully non-interactive zero-knowledge proof of knowledge.
We study the question of basing symmetric key cryptography on weak secrets. In this setting, Alice and Bob share an n-bit secret W, which might not be uniformly random, but the adversary has at least k bits of uncertainty about it (formalized using conditional min-entropy). Since standard symmetric-key primitives require uniformly random secret keys, we would like to construct an authenticated key agreement protocol in which Alice and Bob use W to agree on a nearly uniform key R, by communicating over a public channel controlled by an active adversary Eve. We study this question in the information theoretic setting where the attacker is computationally unbounded. We show that single-round (i.e. one message) protocols do not work when k ≤ n/2, and require poor parameters even when n/2<k<<n.