International Association
for Cryptologic Research

When modelling how public key encryption can enable secure communication, we should acknowledge that secret information, such as private keys or the encryption’s randomness, could become compromised. Intuitively, one would expect unrelated communication to remain secure, yet formalizing this intuition has proven challenging. Several security notions have appeared that aim to capture said scenario, ranging from the multi-user setting with corruptions, via selective opening attacks (SOA), to non-committing encryption (NCE). Remarkably, how the different approaches compare has not yet been systematically explored. We provide a novel framework that maps each approach to an underlying philosophy of confidentiality: indistinguishability versus simulatability based, each with an a priori versus an a posteriori variant, leading to four distinct philosophies. In the absence of corruptions, these notions are largely equivalent; yet, in the presence of corruptions, they fall into a hierarchy of relative strengths, from IND-CPA and IND-CCA at the bottom, via indistinguishability SOA and simulatability SOA, to NCE at the top. We provide a concrete treatment for the four notions, discuss subtleties in their definitions and asymptotic interpretations and identify limitations of each. Furthermore, we re-cast the main implications of the hierarchy in a concrete security framework, summarize and contextualize other known relations, identify open problems, and close a few gaps.

Side channel evaluations benefit from sound characterisations of adversarial leakage models, which are the determining factor for attack success. Two questions are of interest: can we define and estimate a quantity that captures the ideal adversary (who knows all the distributions that are involved in an attack), and can we define and estimate a quantity that captures a concrete adversary (represented by a given leakage model)? Existing work has led to a proliferation of custom quantities to measure both types of adversaries, which can be data intensive to estimate in the ideal case, even for discrete side channels and especially when the number of dimensions in the side channel traces grows. In this paper, we show how to define the mutual information between carefully chosen variables of interest and how to instantiate a recently suggested mutual information estimator for practical estimation. We apply our results to real-world data sets and are the first to provide a mutual information-based characterisation of ideal and concrete adversaries utilising up to 30 data points.

Mass surveillance targets many users at the same time with the goal of learning as much as possible. Intuitively, breaking many users’ cryptography simultaneously should be at least as hard as that of only breaking a single one, but ideally security degradation is gradual: an adversary ought to work harder to break more. Bellare, Ristenpart and Tessaro (Crypto’12) introduced the notion of multi-instance security to capture the related concept for password hashing with salts. Auerbach, Giacon and Kiltz (Eurocrypt’20) motivated the study of public key encryption (PKE) in the multi-instance setting, yet their technical results are exclusively stated in terms of key encapsulation mechanisms (KEMs), leaving a considerable gap. We investigate the multi-instance security of public key encryption. Our contributions are twofold. Firstly, we define and compare possible security notions for multi-instance PKE, where we include PKE schemes whose correctness is not perfect. Secondly, we observe that, in general, a hybrid encryption scheme of a multi-instance secure KEM and an arbitrary data encapsulation mechanism (DEM) is unlikely to inherit the KEM’s multi-instance security. Yet, we show how with a suitable information-theoretic DEM, and a computationally secure key derivation function if need be, inheritance is possible. As far as we are aware, ours is the first inheritance result in the challenging multi-bit scenario.