Section: New Results

Lattices: algorithms and cryptology

Reduction of orthogonal lattice bases

As a typical application, the LLL lattice basis reduction algorithm is applied to bases of the orthogonal lattice of a given integer matrix, via reducing lattice bases of a special type. With such bases in input, we have proposed in [26] a new technique for bounding from above the number of iterations required by the LLL algorithm. The main technical ingredient is a variant of the classical LLL potential, which could prove useful to understand the behavior of LLL for other families of input bases.

Lattice-Based Zero-Knowledge Arguments for Integer Relations

The paper [36] provides lattice-based protocols allowing to prove relations among committed integers. While the most general zero-knowledge proof techniques can handle arithmetic circuits in the lattice setting, adapting them to prove statements over the integers is non-trivial, at least if we want to handle exponentially large integers while working with a polynomial-size modulus q. For a polynomial L, the paper provides zero-knowledge arguments allowing a prover to convince a verifier that committed L-bit bitstrings x, y and z are the binary representations of integers X, Y and Z satisfying Z=X+Y over . The complexity of the new arguments is only linear in L. Using them, the paper constructs arguments allowing to prove inequalities X<Z among committed integers, as well as arguments showing that a committed X belongs to a public interval [α,β], where α and β can be arbitrarily large. The new range arguments have logarithmic cost (i.e., linear in L) in the maximal range magnitude. Using these tools, the paper obtains zero-knowledge arguments showing that a committed element X does not belong to a public set S using O(n·log|S|) bits of communication, where n is the security parameter. The paper finally gives a protocol allowing to argue that committed L-bit integers X, Y and Z satisfy multiplicative relations Z=XY over the integers, with communication cost subquadratic in L. To this end, the paper uses its new protocol for integer addition to prove the correct recursive execution of Karatsuba's multiplication algorithm. The security of the new protocols relies on standard lattice assumptions with polynomial modulus and polynomial approximation factor.

Logarithmic-Size Ring Signatures With Tight Security from the DDH Assumption

Ring signatures make it possible for a signer to anonymously and, yet, convincingly leak a secret by signing a message while concealing his identity within a flexibly chosen ring of users. Unlike group signatures, they do not involve any setup phase or tracing authority. Despite a lot of research efforts in more than 15 years, most of their realizations require linear-size signatures in the cardinality of the ring. In the random oracle model, two recent constructions decreased the signature length to be only logarithmic in the number N of ring members. On the downside, their suffer from rather loose reductions incurred by the use of the Forking Lemma. This paper considers the problem of proving them tightly secure without affecting their space efficiency. Surprisingly, existing techniques for proving tight security in ordinary signature schemes do not trivially extend to the ring signature setting. The paper [37] overcomes these difficulties by combining the Groth-Kohlweiss Σ-protocol (Eurocrypt'15) with dual-mode encryption schemes. The main result is a fully tight construction based on the Decision Diffie-Hellman assumption in the random oracle model. By full tightness, we mean that the reduction's advantage is as large as the adversary's, up to a constant factor.

Adaptively Secure Distributed PRFs from LWE

In distributed pseudorandom functions (DPRFs), a PRF secret key SK is secret shared among N servers so that each server can locally compute a partial evaluation of the PRF on some input X. A combiner that collects t partial evaluations can then reconstruct the evaluation F(SK,X) of the PRF under the initial secret key. So far, all non-interactive constructions in the standard model are based on lattice assumptions. One caveat is that they are only known to be secure in the static corruption setting, where the adversary chooses the servers to corrupt at the very beginning of the game, before any evaluation query. The paper [38] constructs the first fully non-interactive adaptively secure DPRF in the standard model. The construction is proved secure under the LWE assumption against adversaries that may adaptively decide which servers they want to corrupt. The new construction is also extended in order to achieve robustness against malicious adversaries.

Unbounded ABE via Bilinear Entropy Expansion, Revisited

This paper [24] presents simpler and improved constructions of unbounded attribute-based encryption (ABE) schemes with constant-size public parameters under static assumptions in bilinear groups. Concretely, we obtain: a simple and adaptively secure unbounded ABE scheme in composite-order groups, improving upon a previous construction of Lewko and Waters (Eurocrypt’11) which only achieves selective security; an improved adaptively secure unbounded ABE scheme based on the k-linear assumption in prime-order groups with shorter ciphertexts and secret keys than those of Okamoto and Takashima (Asiacrypt’12); the first adaptively secure unbounded ABE scheme for arithmetic branching programs under static assumptions. At the core of all of these constructions is a “bilinear entropy expansion” lemma that allows us to generate any polynomial amount of entropy starting from constant-size public parameters; the entropy can then be used to transform existing adaptively secure “bounded” ABE schemes into unbounded ones.

Improved Anonymous Broadcast Encryptions: Tight Security and Shorter Ciphertext

This paper [35] investigates anonymous broadcast encryptions (ANOBE) in which a ciphertext hides not only the message but also the target recipients associated with it. Following Libert et al.'s generic construction [PKC, 2012], we propose two concrete ANOBE schemes with tight reduction and better space efficiency.

  • -

    The IND-CCA security and anonymity of our two ANOBE schemes can be tightly reduced to standard k-Linear assumption (and the existence of other primitives). For a broadcast system with n users, Libert et al.'s security analysis suffers from 𝒪(n3) loss while our security loss is constant.

  • -

    Our first ANOBE supports fast decryption and has a shorter ciphertext than the fast-decryption version of Libert et al.'s concrete ANOBE. Our second ANOBE is adapted from the first one. We sacrifice the fast decryption feature and achieve shorter ciphertexts than Libert et al.'s concrete ANOBE with the help of bilinear groups. Technically, we start from an instantiation of Libert et al.'s generic ANOBE [PKC, 2012], but we work out all our proofs from scratch instead of relying on their generic security result. This intuitively allows our optimizations in the concrete setting.

Compact IBBE and Fuzzy IBE from Simple Assumptions

This paper [29] proposes new constructions for identity-based broadcast encryption (IBBE) and fuzzy identity-based encryption (FIBE) in composite-order groups equipped with a bilinear pairing. Our starting point is the IBBE scheme of Delerablée (Asiacrypt 2007) and the FIBE scheme of Herranz et al. (PKC 2010) proven secure under parameterized assumptions called generalized decisional bilinear Diffie-Hellman (GDDHE) and augmented multi-sequence of exponents Diffie-Hellman (aMSE-DDH) respectively. The two schemes are described in the prime-order pairing group. We transform the schemes into the setting of (symmetric) composite-order groups and prove security from two static assumptions (subgroup decision). The Déjà Q framework of Chase et al. (Asiacrypt 2016) is known to cover a large class of parameterized assumptions (dubbed "Uber assumption"), that is, these assumptions, when defined in asymmetric composite-order groups, are implied by subgroup decision assumptions in the underlying composite-order groups. We argue that the GDDHE and aMSE-DDH assumptions are not covered by the Déjà Q uber assumption framework. We therefore work out direct security reductions for the two schemes based on subgroup decision assumptions. Furthermore, our proofs involve novel extensions of Déjà Q techniques of Wee (TCC 2016-A) and Chase et al. Our constructions have constant-size ciphertexts. The IBBE has constant-size keys as well and achieves a stronger security guarantee as compared to Delerablée's IBBE, thus making it the first compact IBBE known to be selectively secure without random oracles under simple assumptions. The fuzzy IBE scheme is the first to simultaneously feature constant-size ciphertexts and security under standard assumptions.

Improved Inner-product Encryption with Adaptive Security and Full Attribute-hiding

This paper [25] proposes two IPE schemes achieving both adaptive security and full attribute-hiding in the prime-order bilinear group, which improve upon the unique existing result satisfying both features from Okamoto and Takashima [Eurocrypt'12] in terms of efficiency.

  • -

    Our first IPE scheme is based on the standard k-Lin assumption and has shorter master public key and shorter secret keys than Okamoto and Takashima's IPE under weaker DLIN=2-lin assumption.

  • -

    Our second IPE scheme is adapted from the first one; the security is based on the XDLIN assumption (as Okamoto and Takashima's IPE) but now it also enjoys shorter ciphertexts.

Technically, instead of starting from composite-order IPE and applying existing transformation, we start from an IPE scheme in a very restricted setting but already in the prime-order group, and then gradually upgrade it to our full-fledged IPE scheme. This method allows us to integrate Chen et al.'s framework [Eurocrypt'15] with recent new techniques [TCC'17, Eurocrypt'18] in an optimized way.

Improved Security Proofs in Lattice-Based Cryptography: Using the Rényi Divergence Rather than the Statistical Distance

The Rényi divergence is a measure of closeness of two probability distributions. In this paper [5], we show that it can often be used as an alternative to the statistical distance in security proofs for lattice-based cryptography. Using the Rényi divergence is particularly suited for security proofs of primitives in which the attacker is required to solve a search problem (e.g., forging a signature). We show that it may also be used in the case of distinguishing problems (e.g., semantic security of encryption schemes), when they enjoy a public sampleability property. The techniques lead to security proofs for schemes with smaller parameters, and sometimes to simpler security proofs than the existing ones.

CRYSTALS-Dilithium: A Lattice-Based Digital Signature Scheme

This paper [8] presents Dilithium, a lattice-based signature scheme that is part of the CRYSTALS (Cryptographic Suite for Algebraic Lattices) package that will be submitted to the NIST call for post-quantum standards. The scheme is designed to be simple to securely implement against side-channel attacks and to have comparable efficiency to the currently best lattice-based signature schemes. Our implementation results show that Dilithium is competitive with lattice schemes of the same security level and outperforms digital signature schemes based on other post-quantum assumptions.

On the asymptotic complexity of solving LWE

In this paper [9], we provide for the first time an asymptotic comparison of all known algorithms for the search version of the Learning with Errors (LWE) problem. This includes an analysis of several lattice-based approaches as well as the combinatorial BKW algorithm. Our analysis of the lattice-based approaches defines a general framework, in which the algorithms of Babai, Lindner–Peikert and several pruning strategies appear as special cases. We show that within this framework, all lattice algorithms achieve the same asymptotic complexity. For the BKW algorithm, we present a refined analysis for the case of only a polynomial number of samples via amplification, which allows for a fair comparison with lattice-based approaches. Somewhat surprisingly, such a small number of samples does not make the asymptotic complexity significantly inferior, but only affects the constant in the exponent. As the main result we obtain that both, lattice-based techniques and BKW with a polynomial number of samples, achieve running time 2O(n) for n-dimensional LWE, where we make the constant hidden in the big-O notion explicit as a simple and easy to handle function of all LWE-parameters. In the lattice case this function also depends on the time to compute a BKZ lattice basis with block size Θ(n). Thus, from a theoretical perspective our analysis reveals how LWE ’s complexity changes as a function of the LWE-parameters, and from a practical perspective our analysis is a useful tool to choose LWE-parameters resistant to all currently known attacks.

Measuring, Simulating and Exploiting the Head Concavity Phenomenon in BKZ

The Blockwise-Korkine-Zolotarev (BKZ) lattice reduction algorithm is central in cryptanalysis, in particular for lattice-based cryptography. A precise understanding of its practical behavior in terms of run-time and output quality is necessary for parameter selection in cryptographic design. As the provable worst-case bounds poorly reflect the practical behavior, cryptanalysts rely instead on the heuristic BKZ simulator of Chen and Nguyen (Asiacrypt'11). It fits better with practical experiments, but not entirely. In particular, it over-estimates the norm of the first few vectors in the output basis. Put differently, BKZ performs better than its Chen-Nguyen simulation.

In this article [15], we first report experiments providing more insight on this shorter-than-expected phenomenon. We then propose a refined BKZ simulator by taking the distribution of short vectors in random lattices into consideration. We report experiments suggesting that this refined simulator more accurately predicts the concrete behavior of BKZ. Furthermore, we design a new BKZ variant that exploits the shorter-than-expected phenomenon. For the same cost assigned to the underlying SVP-solver, the new BKZ variant produces bases of better quality. We further illustrate its potential impact by testing it on the SVP-120 instance of the Darmstadt lattice challenge.

CRYSTALS - Kyber: A CCA-Secure Module-Lattice-Based KEM

Rapid advances in quantum computing, together with the announcement by the National Institute of Standards and Technology (NIST) to define new standards for digital signature, encryption, and key-establishment protocols, have created significant interest in post-quantum cryptographic schemes. This paper [17] introduces Kyber (part of CRYSTALS - Cryptographic Suite for Algebraic Lattices - a package submitted to NIST post-quantum standardization effort in November 2017), a portfolio of post-quantum cryptographic primitives built around a key-encapsulation mechanism (KEM), based on hardness assumptions over module lattices. Our KEM is most naturally seen as a successor to the NEWHOPE KEM (Usenix 2016). In particular, the key and ciphertext sizes of our new construction are about half the size, the KEM offers CCA instead of only passive security, the security is based on a more general (and flexible) lattice problem, and our optimized implementation results in essentially the same running time as the aforementioned scheme. We first introduce a CPA-secure public-key encryption scheme, apply a variant of the Fujisaki-Okamoto transform to create a CCA-secure KEM, and eventually construct, in a black-box manner, CCA-secure encryption, key exchange, and authenticated-key-exchange schemes. The security of our primitives is based on the hardness of Module-LWE in the classical and quantum random oracle models, and our concrete parameters conservatively target more than 128 bits of postquantum security.

Learning with Errors and Extrapolated Dihedral Cosets

The hardness of the learning with errors (LWE) problem is one of the most fruitful resources of modern cryptography. In particular, it is one of the most prominent candidates for secure post-quantum cryptography. Understanding its quantum complexity is therefore an important goal. In this paper [20], we show that under quantum polynomial time reductions, LWE is equivalent to a relaxed version of the dihedral coset problem (DCP), which we call extrapolated DCP (eDCP). The extent of extrapolation varies with the LWE noise rate. By considering different extents of extrapolation, our result generalizes Regev's famous proof that if DCP is in BQP (quantum poly-time) then so is LWE (FOCS'02). We also discuss a connection between eDCP and Childs and Van Dam's algorithm for generalized hidden shift problems (SODA'07). Our result implies that a BQP solution for LWE might not require the full power of solving DCP, but rather only a solution for its relaxed version, eDCP, which could be easier.

Pairing-friendly twisted Hessian curves

This paper [27] presents efficient formulas to compute Miller doubling and Miller addition utilizing degree-3 twists on curves with j-invariant 0 written in Hessian form. We give the formulas for both odd and even embedding degrees and for pairings on both G1×G2 and G2×G1. We propose the use of embedding degrees 15 and 21 for 128-bit and 192-bit security respectively in light of the NFS attacks and their variants. We give a comprehensive comparison with other curve models; our formulas give the fastest known pairing computation for embedding degrees 15, 21, and 24.

On the Statistical Leak of the GGH13 Multilinear Mapand some Variants

At EUROCRYPT 2013, Garg, Gentry and Halevi proposed a candidate construction (later referred as GGH13) of cryptographic multilinear map (MMap). Despite weaknesses uncovered by Hu and Jia (EUROCRYPT 2016), this candidate is still used for designing obfuscators. The naive version of the GGH13 scheme was deemed susceptible to averaging attacks, i.e., it could suffer from a statistical leak (yet no precise attack was described). A variant was therefore devised, but it remains heuristic. Recently, to obtain MMaps with low noise and modulus, two variants of this countermeasure were developed by Döttling et al. (EPRINT:2016/599). In this work [28], we propose a systematic study of this statistical leak for all these GGH13 variants. In particular, we confirm the weakness of the naive version of GGH13. We also show that, among the two variants proposed by Döttling et al., the so-called conservative method is not so effective: it leaks the same value as the unprotected method. Luckily, the leak is more noisy than in the unprotected method, making the straightforward attack unsuccessful. Additionally, we note that all the other methods also leak values correlated with secrets. As a conclusion, we propose yet another countermeasure, for which this leak is made unrelated to all secrets. On our way, we also make explicit and tighten the hidden exponents in the size of the parameters, as an effort to assess and improve the efficiency of MMaps.

Higher dimensional sieving for the number field sieve algorithms

Since 2016 and the introduction of the exTNFS (extended tower number field sieve) algorithm, the security of cryptosystems based on nonprime finite fields, mainly the pairing- and torus-based ones, is being reassessed. The feasibility of the relation collection, a crucial step of the NFS variants, is especially investigated. It usually involves polynomials of degree 1, i.e., a search space of dimension 2. However, exTNFS uses bivariate polynomials of at least four coefficients. If sieving in dimension 2 is well described in the literature, sieving in higher dimensions has received significantly less attention. In this work [30], we describe and analyze three different generic algorithms to sieve in any dimension for the NFS algorithms. Our implementation shows the practicability of dimension-4 sieving, but the hardness of dimension-6 sieving.

Speed-Ups and Time-Memory Trade-Offs for Tuple Lattice Sieving

In this work [31], we study speed-ups and time–space trade-offs for solving the shortest vector problem (SVP) on Euclidean lattices based on tuple lattice sieving. Our results extend and improve upon previous work of Bai–Laarhoven–Stehlé [ANTS’16] and Herold–Kirshanova [PKC’17], with better complexities for arbitrary tuple sizes and offering tunable time–memory tradeoffs. The trade-offs we obtain stem from the generalization and combination of two algorithmic techniques: the configuration framework introduced by Herold–Kirshanova, and the spherical locality-sensitive filters of Becker–Ducas–Gama–Laarhoven [SODA’16]. When the available memory scales quasi-linearly with the list size, we show that with triple sieving we can solve SVP in dimension n in time 20.3588n+o(n) and space 20.1887n+o(n), improving upon the previous best triple sieve time complexity of 20.3717n+o(n) of Herold–Kirshanova. Using more memory we obtain better asymptotic time complexities. For instance, we obtain a triple sieve requiring only 20.3300n+o(n) time and 20.2075n+o(n) memory to solve SVP in dimension n. This improves upon the best double Gauss sieve of Becker–Ducas–Gama–Laarhoven, which runs in 20.3685n+o(n) time when using the same amount of space.

Improved Quantum Information Set Decoding

In this paper [34], we present quantum information set decoding (ISD) algorithms for binary linear codes. First, we refine the analysis of the quantum walk based algorithms proposed by Kachigar and Tillich (PQCrypto’17). This refinement allows us to improve the running time of quantum decoding in the leading order term: for an n-dimensional binary linear code the complexity of May-Meurer-Thomae ISD algorithm (Asiacrypt’11) drops down from 20.05904n+o(n) to 20.05806n+o(n). Similar improvement is achieved for our quantum version of Becker-JeuxMay-Meurer (Eurocrypt’12) decoding algorithm. Second, we translate May-Ozerov Near Neighbour technique (Eurocrypt’15) to an ‘updateand-query’ language more common in a similarity search literature. This re-interpretation allows us to combine Near Neighbour search with the quantum walk framework and use both techniques to improve a quantum version of Dumer’s ISD algorithm: the running time goes down from 20.059962n+o(n) to 20.059450+o(n).

Quantum Attacks against Indistinguishablility Obfuscators Proved Secure in the Weak Multilinear Map Model

We present in [39] a quantum polynomial time attack against the GMMSSZ branching program obfuscator of Garg et al. (TCC’16), when instantiated with the GGH13 multilinear map of Garg et al. (EUROCRYPT’13). This candidate obfuscator was proved secure in the weak multilinear map model introduced by Miles et al. (CRYPTO’16). Our attack uses the short principal ideal solver of Cramer et al. (EUROCRYPT’16), to recover a secret element of the GGH13 multilinear map in quantum polynomial time. We then use this secret element to mount a (classical) polynomial time mixed-input attack against the GMMSSZ obfuscator. The main result of this article can hence be seen as a classical reduction from the security of the GMMSSZ obfuscator to the short principal ideal problem (the quantum setting is then only used to solve this problem in polynomial time). As an additional contribution, we explain how the same ideas can be adapted to mount a quantum polynomial time attack against the DGGMM obfuscator of Döttling et al. (ePrint 2016), which was also proved secure in the weak multilinear map model.

On the Ring-LWE and Polynomial-LWE Problems

The Ring Learning With Errors problem (RLWE) comes in various forms. Vanilla RLWE is the decision dual-RLWE variant, consisting in distinguishing from uniform a distribution depending on a secret belonging to the dual OK of the ring of integers OK of a specified number field K. In primal-RLWE, the secret instead belongs to OK. Both decision dual-RLWE and primal-RLWE enjoy search counterparts. Also widely used is (search/decision) Polynomial Learning With Errors (PLWE), which is not defined using a ring of integers OK of a number field K but a polynomial ring Z[x]/f for a monic irreducible fZ[x]. We show that there exist reductions between all of these six problems that incur limited parameter losses. More precisely: we prove that the (decision/search) dual to primal reduction from Lyubashevsky et al. [EUROCRYPT 2010] and Peikert [SCN 2016] can be implemented with a small error rate growth for all rings (the resulting reduction is nonuniform polynomial time); we extend it to polynomial-time reductions between (decision/search) primal RLWE and PLWE that work for a family of polynomials f that is exponentially large as a function of deg(f) (the resulting reduction is also non-uniform polynomial time); and we exploit the recent technique from Peikert et al. [STOC 2017] to obtain a search to decision reduction for RLWE for arbitrary number fields. The reductions incur error rate increases that depend on intrinsic quantities related to K and f.

Non-Trivial Witness Encryption and Null-iO from Standard Assumptions

A witness encryption (WE) scheme can take any NP statement as a public-key and use it to encrypt a message. If the statement is true then it is possible to decrypt the message given a corresponding witness, but if the statement is false then the message is computationally hidden. Ideally, the encryption procedure should run in polynomial time, but it is also meaningful to define a weaker notion, which we call non-trivially exponentially efficient WE (XWE), where the encryption run-time is only required to be much smaller than the trivial 2m bound for NP relations with witness size m. In [19], we show how to construct such XWE schemes for all of NP with encryption run-time 2m/2 under the sub-exponential learning with errors (LWE) assumption. For NP relations that can be verified in 𝖭𝖢1 (e.g., SAT) we can also construct such XWE schemes under the sub-exponential Decisional Bilinear Diffie-Hellman (DBDH) assumption. Although we find the result surprising, it follows via a very simple connection to attribute-based encryption.

We also show how to upgrade the above results to get non-trivially exponentially efficient indistinguishability obfuscation for null circuits (niO), which guarantees that the obfuscations of any two circuits that always output 0 are indistinguishable. In particular, under the LWE assumptions we get a XniO scheme where the obfuscation time is 2n/2 for all circuits with input size n. It is known that in the case of indistinguishability obfuscation (iO) for all circuits, non-trivially efficient XiO schemes imply fully efficient iO schemes (Lin et al., PKC 2016) but it remains as a fascinating open problem whether any such connection exists for WE or niO.

Lastly, we explore a potential approach toward constructing fully efficient WE and niO schemes via multi-input ABE.

Function-Revealing Encryption

Multi-input functional encryption is a paradigm that allows an authorized user to compute a certain function—and nothing more—over multiple plaintexts given only their encryption. The particular case of two-input functional encryption has very exciting applications, including comparing the relative order of two plaintexts from their encrypted form (order-revealing encryption).

While being extensively studied, multi-input functional encryption is not ready for a practical deployment, mainly for two reasons. First, known constructions rely on heavy cryptographic tools such as multilinear maps. Second, their security is still very uncertain, as revealed by recent devastating attacks.

In [33], we investigate a simpler approach towards obtaining practical schemes for functions of particular interest. We introduce the notion of function-revealing encryption, a generalization of order-revealing encryption to any multi-input function as well as a relaxation of multi-input functional encryption. We then propose a simple construction of order-revealing encryption based on function-revealing encryption for simple functions, namely orthogonality testing and intersection cardinality. Our main result is an efficient order-revealing encryption scheme with limited leakage based on the standard DLin assumption.

Exploring Crypto Dark Matter: New Simple PRF Candidates and Their Applications

Pseudorandom functions (PRFs) are one of the fundamental building blocks in cryptography. We explore a new space of plausible PRF candidates that are obtained by mixing linear functions over different small moduli. Our candidates are motivated by the goals of maximizing simplicity and minimizing complexity measures that are relevant to cryptographic applications such as secure multiparty computation.

In [16], we present several concrete new PRF candidates that follow the above approach. Our main candidate is a weak PRF candidate (whose conjectured pseudorandomness only holds for uniformly random inputs) that first applies a secret mod-2 linear mapping to the input, and then a public mod-3 linear mapping to the result. This candidate can be implemented by depth-2 𝖠𝖢𝖢0 circuits. We also put forward a similar depth-3 strong PRF candidate. Finally, we present a different weak PRF candidate that can be viewed as a deterministic variant of “Learning Parity with Noise” (LPN) where the noise is obtained via a mod-3 inner product of the input and the key.

The advantage of our approach is twofold. On the theoretical side, the simplicity of our candidates enables us to draw natural connections between their hardness and questions in complexity theory or learning theory (e.g., learnability of depth-2 ACC0 circuits and width-3 branching programs, interpolation and property testing for sparse polynomials, and natural proof barriers for showing super-linear circuit lower bounds). On the applied side, the “piecewise-linear” structure of our candidates lends itself nicely to applications in secure multiparty computation (MPC). Using our PRF candidates, we construct protocols for distributed PRF evaluation that achieve better round complexity and/or communication complexity (often both) compared to protocols obtained by combining standard MPC protocols with PRFs like AES, LowMC, or Rasta (the latter two are specialized MPC-friendly PRFs). Our advantage over competing approaches is maximized in the setting of MPC with an honest majority, or alternatively, MPC with preprocessing.

Finally, we introduce a new primitive we call an encoded-input PRF, which can be viewed as an interpolation between weak PRFs and standard (strong) PRFs. As we demonstrate, an encoded-input PRF can often be used as a drop-in replacement for a strong PRF, combining the efficiency benefits of weak PRFs and the security benefits of strong PRFs. We conclude by showing that our main weak PRF candidate can plausibly be boosted to an encoded-input PRF by leveraging error-correcting codes.

Related-Key Security for Pseudorandom Functions Beyond the Linear Barrier

Related-key attacks (RKAs) concern the security of cryptographic primitives in the situation where the key can be manipulated by the adversary. In the RKA setting, the adversary’s power is expressed through the class of related-key deriving (RKD) functions which the adversary is restricted to using when modifying keys. Bellare and Kohno (Eurocrypt 2003) first formalised RKAs and pin-pointed the foundational problem of constructing RKA-secure pseudorandom functions (RKA-PRFs). To date there are few constructions for RKA-PRFs under standard assumptions, and it is a major open problem to construct RKA-PRFs for larger classes of RKD functions. We make significant progress on this problem. In [3], we first show how to repair the Bellare-Cash framework for constructing RKA-PRFs and extend it to handle the more challenging case of classes of RKD functions that contain claws. We apply this extension to show that a variant of the NaorReingold function already considered by Bellare and Cash is an RKA-PRF for a class of affine RKD functions under the DDH assumption, albeit with an exponential-time security reduction. We then develop a second extension of the Bellare-Cash framework, and use it to show that the same Naor-Reingold variant is actually an RKA-PRF for a class of degree d polynomial RKD functions under the stronger decisional d-Diffie-Hellman inversion assumption. As a significant technical contribution, our proof of this result avoids the exponential-time security reduction that was inherent in the work of Bellare and Cash and in our first result.

Practical Fully Secure Unrestricted Inner Product Functional Encryption modulo p

In [23], we provide adaptively secure functional encryption schemes for the inner product functionality which are both efficient and allow for the evaluation of unbounded inner products modulo a prime p. Our constructions rely on new natural cryptographic assumptions in a cyclic group containing a subgroup where the discrete logarithm (DL) problem is easy which extend Castagnos and Laguillaumie's assumption (RSA 2015) of a DDH group with an easy DL subgroup. Instantiating our generic construction using class groups of imaginary quadratic fields gives rise to the most efficient functional encryption for inner products modulo an arbitrary large prime p. One of our schemes outperforms the DCR variant of Agrawal et al.'s protocols in terms of size of keys and ciphertexts by factors varying between 2 and 20 for a 112-bit security.