In this work we challenge the common misconception that information-theoretic (IT) privacy is too impractical to be used in the real-world: we propose to build simple and reusable IT-encryption solutions whose only efficiency penalty (compared to computationally-secure schemes) comes from a large secret key size, which is often a rather minor inconvenience, as storage is cheap. In particular, our solutions are stateless and locally computable at the optimal rate, meaning that honest parties do not maintain state and read only (optimally) small portions of their large keys with every use. Moreover, we also propose a novel architecture for outsourcing the storage of these long keys to a network of semi-trusted servers, trading the need to store large secrets with the assumption that it is hard to simultaneously compromise too many publicly accessible ad-hoc servers. Our architecture supports everlasting privacy and post-application security of the derived one-time keys, resolving two major limitations of a related model for outsourcing key storage, called bounded storage model. Both of these results come from nearly optimal constructions of so called doubly-affine extractors: locally-computable, seeded extractors Ext(X,S) which are linear functions of X (for any fixed seed S), and protect against bounded affine leakage on X. This holds unconditionally, even if (a) affine leakage may adaptively depend on the extracted key R = Ext(X,S); and (b) the seed S is only computationally secure. Neither of these properties are possible with general-leakage extractors.
In this work we challenge the common misconception that information-theoretic (IT) privacy is too impractical to be used in the real-world: we propose to build simple and reusable IT-encryption solutions whose only efficiency penalty (compared to computationally-secure schemes) comes from a large secret key size, which is often a rather minor inconvenience, as storage is cheap. In particular, our solutions are stateless and locally computable at the optimal rate, meaning that honest parties do not maintain state and read only (optimally) small portions of their large keys with every use. Moreover, we also propose a novel architecture for outsourcing the storage of these long keys to a network of semi-trusted servers, trading the need to store large secrets with the assumption that it is hard to simultaneously compromise too many publicly accessible ad-hoc servers. Our architecture supports everlasting privacy and post-application security of the derived one-time keys, resolving two major limitations of a related model for outsourcing key storage, called bounded storage model. Both of these results come from nearly optimal constructions of so called doubly-affine extractors: locally-computable, seeded extractors Ext(X,S) which are linear functions of X (for any fixed seed S), and protect against bounded affine leakage on X. This holds unconditionally, even if (a) affine leakage may adaptively depend on the extracted key R = Ext(X,S); and (b) the seed S is only computationally secure. Neither of these properties are possible with general-leakage extractors.
We address the problem of the-security of cryptographic protocols in face of future advances in computing technology and algorithmic research. The problem stems from the fact may be deemed that computations which at a given point in time may be deemed infeasible, can, in the course of years or decades, be made possible with improved hardware and/or breakthroughs in code-breaking algorithms. In such cases, the security of historical , but nonetheless highly confidential data may be in jeopardy. We present a scheme for efficient secure two-party communication with provable everlasting security. The security is guaranteed in face of any future technological advances, given the current state of of the art. Furthermore, the security of the messages is also guaranteed even if the secret encryption/decryption key is revealed in the future, The scheme is based on the bounded storage model and provides information-theoretic security in this model. The bounded storage model postulates an adversary who is computationally unbounded, and is only bounded in the amount of storage (not computation space) available to store the output of his computation. The bound on the storage can be arbitrarily large (e.g., 100 Tbytes), as long as it is fixed. Given this storage bound, our protocols guarantee that even a computationally all powerful adversary gains no information about a message (except with a probability that is exponentially small in the security parameter k). The bound on storage space need only hold at the time of the message transmission. Thereafter, no additional storage space or, computational power can help the adversary in deciphering the message. We present two protocols. The first protocol, which elaborates on the autoregressive (AR) protocol of Aumann and Rabin (see Advances in Cryptology-Crypto '99, p. 65-79, 1999), employs a short secret key whose size is independent of the length of the message, but uses many public random bits. The second protocol uses an optimal number of public random bits, but employs a longer secret key. Our proof of security utilizes a novel linear algebraic technique.
We study the classical problem of privacy amplification, where two parties Alice and Bob share a weak secret X of min-entropy k, and wish to agree on secret key R of length m over a public communication channel completely controlled by a computationally unbounded attacker Eve. Despite being extensively studied in the literature, the problem of designing “optimal ” efficient privacy amplification protocols is still open, because there are several optimization goals. The first of them is (1) minimizing the entropy loss L = k−m (it is known that the optimal value for L = O(λ), where ε = 2−λ is the desired security of the protocol). Other important considerations include (2) minimizing the number of communication rounds, (3) maintaining security even after the secret key is used (this is called post-application robustness), and (4) ensuring that the protocol P does not leak some “useful information ” about the source X (this is called source privacy). Additionally, when dealing with a very long source X, as happens in the so-called Bounded Retrieval Model (BRM), extracting as long a key as possible is no longer the goal. Instead, the goals are (5) to touch as little of X as possible (for efficiency), and (6) to be able to run the protocol many times on the same X, extracting multiple secure keys.
Motivated by applications in cryptography, we consider a generalization of randomness extraction and the related notion of privacy amplification to the case of two correlated sources. We introduce the notion of correlation extractors, which extract nearly perfect independent instances of a given joint distribution from imperfect, or "leaky," instances of the same distribution. More concretely, suppose that Alice holds a and Bob holds b, where (a, b) are obtained by taking n independent samples from a joint distribution (X, Y) and letting a include all X instances and b include all Y instances. An adversary Eve obtains partial information about (a, b) by choosing a function L with output length t and learning L(a, b). The goal is to design a protocol between Alice and Bob which may use additional fresh randomness, such that for every L as above the following holds. In the end of the interaction, Alice outputs a' and Bob outputs b' such that (a', b') are statistically indistinguishable from m independent instances of (X, Y) even when conditioned on Eve's view, and even when conditioned on the joint view of Eve together with either Alice or Bob. The standard questions of privacy amplification and randomness extraction correspond to the case where X and Y are identical random bits. In this work we address this question for other types of correlations. A central special case is that of OT extractors, which are correlation extractors for the correlation (X, Y) corresponding to the cryptographic primitive of oblivious transfer. Our main result is that for any finite joint distribution (X, Y) there is an explicit correlation extractor which extracts m = ?(n) instances using O(n) bits of communication, even when t = ?(n) bits of information can be leaked to Eve. We present several applications which motivate the concept of correlation extractors and our main result. These include: ? Protecting certain cryptographic protocols against sidechannel attacks. ? A protocol which realizes m instances of oblivious transfer by communicating only O(m) bits. The security of the protocol relies on a number-theoretic intractability assumption. ? A constant-rate unconditionally secure construction of oblivious transfer (for semi-honest parties) from any nontrivial channel. This establishes constant-rate equivalence of any two nontrivial finite channels.
In the bounded-storage model for information-theoretically secure encryption and key-agreement one can prove the security of a cipher based on the sole assumption that the adversary's storage capacity is bounded, say by s bits, even if her computational power is unlimited. Assume that a random t-bit string R is either publicly available (e.g. the signal of a deep space radio source) or broadcast by one of the legitimate parties. If s < t, the adversary can store only partial information about R. The legitimate sender Alice and receiver Bob, sharing a short secret key K initially, can therefore potentially generate a very long n-bit one-time pad X with n jKj about which the adversary has essentially no information, thus at rst glance apparently contradicting Shannon's bound on the key size of a perfect cipher.
We study the design of cryptographic primitives resilient to key-leakage attacks, where an attacker can repeatedly and adaptively learn information about the secret key, subject only to the constraint that the overall amount of such information is bounded by some parameter ℓ. We construct a variety of leakage-resilient public-key systems including the first known identification schemes (ID), signature schemes and authenticated key agreement protocols (AKA). Our main result is an efficient three-round leakage-resilient AKA in the Random-Oracle model. This protocol ensures that session keys are private and authentic even if (1) the adversary leaks a large fraction of the long-term secret keys of both users prior to the protocol execution and (2) the adversary completely learns the long-term secret keys after the protocol execution. In particular, our AKA protocol provides qualitatively stronger privacy guarantees than leakage-resilient public-encryption schemes (constructed in prior and concurrent works), since such schemes necessarily become insecure if the adversary can perform leakage attacks after seing a ciphertext. Moreover, our schemes can be flexibly extended to the Bounded-Retrieval Model, allowing us to tolerate very large absolute amount of adversarial leakage ℓ (potentially many gigabytes of
Everlasting (EL) privacy offers an attractive solution to the Store-Now-Decrypt-Later (SNDL) problem, where future increases in the attacker's capability could break systems which are believed to be secure today. Instead of requiring full information-theoretic security, everlasting privacy allows computationally-secure transmissions of ephemeral secrets, which are only "effective" for a limited periods of time, after which their compromise is provably useless for the SNDL attacker. In this work we revisit such everlasting privacy model of Dodis and Yeo (ITC'21), which we call Hypervisor EverLasting Privacy (HELP). HELP is a novel architecture for generating shared randomness using a network of semi-trusted servers (or "hypervisors"), trading the need to store/distribute large shared secrets with the assumptions that it is hard to: (a) simultaneously compromise too many publicly accessible ad-hoc servers; and (b) break a computationally-secure encryption scheme very quickly. While Dodis and Yeo presented good HELP solutions in the asymptotic sense, their solutions were concretely expensive and used heavy tools (like large finite fields or gigantic Toeplitz matrices). We abstract and generalize the HELP architecture to allow for more efficient instantiations, and construct several concretely efficient HELP solutions. Our solutions use elementary cryptographic operations, such as hashing and message authentication. We also prove a very strong composition theorem showing that our EL architecture can use any message transmission method which is computationally-secure in the Universal Composability (UC) framework. This is the first positive composition result for everlasting privacy, which was otherwise known to suffer from many "non-composition" results (Müller-Quade and Unruh; J of Cryptology'10).