Search

At present, with the popularity of Internet of things (IoT), a huge number of datasets generated by IoT devices are being uploaded to the cloud storage in remote data management service, but a series of security and privacy defects also arises, where one of the best ways for preventing data disclosure is encryption. Among them, searchable encryption (SE) is considered to be a very attractive cryptographic technology, since it allows users to search records in an encrypted form and to protect user’s data on an untrusted server. For the sake of enhancing search permission, attribute-based keyword search (ABKS) is an efficient method to provide secure search queries and fine-grained access authentications over ciphertexts. However, most existing ABKS schemes concentrate on single keyword search, which usually returns redundant and irrelevant results, so it would cost some unnecessary computation and communication resources. Furthermore, existing work in the literature mostly only supports unshared multiowner where a specific data owner owns each file, which is not able to satisfy more desired expressive search. In this work, we propose a novel attribute-based multikeyword search for shared multiowner (ABMKS-SM) primitive in IoT to achieve enhanced access control for users; meanwhile, it can support multikeyword search over ciphertexts and give a formal security analysis in the adaptive against chosen keyword attack (IND-CKA) model. Finally, we have also implemented this prototype to show efficiency when compared with some previous schemes.

Since Kocher (CRYPTO’96) proposed timing attack, side channel analysis (SCA) has shown great potential to break cryptosystems via physical leakage. Recently, deep learning techniques are widely used in SCA and show equivalent and even better performance compared to traditional methods. However, it remains unknown why and when deep learning techniques are effective and efficient for SCA. Masure et al. (IACR TCHES 2020(1):348–375) illustrated that deep learning paradigm is suitable for evaluating implementations against SCA from a worst-case scenario point of view, yet their work is limited to balanced data and a specific loss function. Besides, deep learning metrics are not consistent with side channel metrics. In most cases, they are deceptive in foreseeing the feasibility and complexity of mounting a successful attack, especially for imbalanced data. To mitigate the gap between deep learning metrics and side channel metrics, we propose a novel Cross Entropy Ratio (CER) metric to evaluate the performance of deep learning models for SCA. CER is closely related to traditional side channel metrics Guessing Entropy (GE) and Success Rate (SR) and fits to deep learning scenario. Besides, we show that it works stably while deep learning metrics such as accuracy becomes rather unreliable when the training data tends to be imbalanced. However, estimating CER can be done as easy as natural metrics in deep learning algorithms with low computational complexity. Furthermore, we adapt CER metric to a new kind of loss function, namely CER loss function, designed specifically for deep learning in side channel scenario. In this way, we link directly the SCA objective to deep learning optimization. Our experiments on several datasets show that, for SCA with imbalanced data, CER loss function outperforms Cross Entropy loss function in various conditions., IACR Transactions on Cryptographic Hardware and Embedded Systems, Volume 2020, Issue 3

In cyber security, authenticated key exchange (AKE) can be used to achieve the privacy and authentication of data. As a relevant cryptographic protocol, password-based authenticated key exchange (PAKE) has been studied for its convenience. Recently, Katz and Vaikuntanathan proposed a round-optimal PAKE from smooth projective hash functions (SPHFs). However, the instantiation of smooth projective hash functions depends on the underlying NP-relation which is a CCA-secure encryption relation in their construction. In this paper, we apply a new cryptographic primitive named witness PRFs to construct PAKE. In our settings, the concrete construction of witness PRFs is independent of the underlying NP-relation. At this point, our construction is more general, and furthermore, we have a discussion on some possible NP-relations, which could be used to construct secure PAKE in our settings.

Pseudorandom functions (PRFs) serve as a fundamental cryptographic primitive that is essential for encryption, identification and authentication. The concept of PRFs first formalized by Goldreich, Goldwasser, and Micali (JACM 1986), and their construction is based on length-doubling pseudorandom generators (PRGs) by using the tree-extention technique. Subsequently, Naor and Reingold proposed a construction based on synthesizers (JACM 2004) which can be instantiated from factoring and the Diffie-Hellman assumption. Recently, some efficient constructions were proposed in the post-quantum background. Banerjee, Peikert, and Rosen (Eurocrypt 2012) constructed relatively more efficient PRFs based on “learning with error” (LWE). Soon afterwards, Yu and Steinberger (Eurocrypt 2016) proposed two efficient constructions of randomized PRFs (with public coin as a parameter) from “learning parity with noise” (LPN). In this paper, we construct standard and randomized PRFs via Mersenne prime assumptions which were proposed by Aggarwal et al. (Crypto 2018) as new post-quantum candidate hardness assumptions. In contrast with Yu and Steinberger’s constructions, our first construction could have the same parameters to their second construction but not needs extra public coin and our second construction has a smaller public coin and key size comparing with their first construction.