Database as a service (DBaaS) in cloud raises severe concern in terms of data security. Storing data in encrypted form may confirm data confidentiality. But, database processing cannot be supported in this encrypted form. Theoretically, homomorphic encryption is a solution to support direct encrypted data processing. In this work, we highlight one of the major challenges of FHE- encrypted query processing that demands huge data transfer requirements from cloud to client for final decryption at the end of SQL query execution. We show in light of Chosen Plaintext Attack (CPA) that in spite of performing conditional filtering through SQL queries over encrypted databases, the size of the resultant dataset cannot be less than the original size of the database. In this work, we make an effort to propose a new encrypted query processing framework termed asSHELDBwhich supports client storage compatibility and low communication overhead using block-wise final result transmission from cloud to client by extending the concept of TOP operator implementation in standard SQL. However, it is to be noted that realization of such optimization is not straightforward because of circuit-based implementation requirements with FHE gates. Our experimental demonstration shows that the proposed framework is capable of executing all TPC-C standard SQL queries with the aid of 8-core parallel processing within$\sim 12.65$minutes for an encrypted database of$768 \times 9$size with 16-bits elements each. Though the computation time is linear with the number of rows, we have explored map-reduce type parallel processing techniques to reduce the timing requirements for databases with larger rows. Consequently, our new query processing framework reduces the communication overhead from m to$\delta k$rows ($1 \leq \delta \leq block$) where the encrypted database contains m rows,$\delta$is the number of blocks to be transmitted each time with$k$= ($m/block$) rows. In spite of$k$being a controllable parameter according to client storage and$\delta$is dependent on the query parameters, final security analysis explains why the proposed technique is general database attack-resistant.

Fully Homomorphic Encryption (FHE) schemes are widely used cryptographic primitives for performing arbitrary computations on encrypted data. However, FHE incorporates a computationally intensive mechanism called bootstrapping, that resets the noise in the ciphertext to a lower level allowing the computation on circuits of arbitrary depth. This process can take significant time, ranging from several minutes to hours. To address the above issue, in this work, we propose an Electronic Design Automation (EDA) framework$\mathsf{FHEDA}$that generates efficient Boolean representations of circuits compatible with the Torus-FHE (ASIACRYPT 2020) scheme. To the best of our knowledge, this is the first work in the EDA domain of FHE. We integrate logic synthesis and gate optimization techniques into our$\mathsf{FHEDA}$framework for reducing the total number of bootstrapping operations in a Boolean circuit, which leads to a significant (up to 50%) reduction in homomorphic computation time. Our$\mathsf{FHEDA}$is built upon the observation that in Torus-FHE two consecutive Boolean gate evaluations over fresh encryptions require only one bootstrapping instead of two, based on appropriate parameter choices. By integrating this observation with logic replacement techniques into$\mathsf{FHEDA}$, we could reduce the total number of bootstrapping operations along with the circuit depth. This eventually reduces the homomorphic evaluation time of Boolean circuits. In order to verify the efficacy of our approach, we assess the performance of the proposed EDA flow on a diverse set of representative benchmarks including privacy-preserving machine learning and different symmetric key block ciphers.

In public and shared platforms where security is paramount, encrypted database storage and processing have become a key research priority. However, traditionally encrypted databases do not support direct computation of relational queries, and hence, those types of query processing are infeasible without intermediate decryption. To achieve end-to-end encrypted computation, such databases prefer fully homomorphic encryption (FHE), which demands circuit-based representation of any algorithm. In our work, along with other encrypted SQL operations, we analyze SQL join and show straightforward implementation, which incurs huge performance overhead. Not only in performance, number of ciphertext expansions is also huge in this case, which can trigger a sharp rise in performance overhead in encrypted domain. To address these challenges, we propose a lightweight solution only for joining the encrypted databases using pair-wise traditional column encryption with the introduction of primary and foreign keys. Then, we highlight how to perform homomorphic operations over that encrypted joining result. Our implementation shows 89.34% improvement compared to straightforward FHE-joining. Consequently, we demonstrate memory overhead reduction ∼13%.

Cloud Computing is a promising solution in distributed internet computing for IT and scientific research. However, data storage and computing in the cloud domain raise added questions in terms of security. Data encrypted with traditional encryption schemes may confirm confidentiality, but computation in cloud domain becomes infeasible. This paper focuses on designing an encrypted database considering homomorphic encryption (HE) as an underlying scheme so that query execution is carried out on encrypted version of the database without any need of intermediate decryption. Existing encrypted databases are either based on partial HE or deterministic HE to achieve practical performance; hence, they are either limited in terms of types of query execution or prone to known attacks. To mitigate these issues, we explore the practical challenges of a fully homomorphic encryption (FHE)-based database design. FHE theoretically promises to perform arbitrary operations on encrypted data. However, realizing any algorithm in homomorphic domain requires a circuit-based representation of that specific algorithm, which is a non-trivial task. In this work, we explore the practical challenges of FHE database design, mostly in the case of multi-user organizational scenarios, and propose a scheme for secure modification or conditional update of the encrypted database. Moreover, when an organization outsources a database to the cloud, a single FHE key is used to encrypt all columns of the database. However, all users(or employees) of the same organization should not have equal read and write access permission to the whole database. In this work, we propose an architecture to apply Attribute-Based Access Control (ABAC) on FHE databases with minimum overhead in terms of performance and storage. We propose required changes in registration, login, and user revocation phases for our scheme to perform conditional SQL query processing on FHE encrypted database (EDB). Our proposed framework is capable of performing end-to-end encrypted conditional UPDATE with suitable access control within 17 minutes on a multi-core processor platform for 769 rows and 9 columns of database size with 16-bit size of data. To the best of our knowledge, our proposed technique is the first one in literature to support arbitrary secure encrypted SQL query execution with suitable access control.

Homomorphic encryption (HE) is the ultimate tool for performing secure computations even in untrusted environments. Application of HE for deep learning (DL) inference is an active area of research, given the fact that DL models are often deployed in untrusted environments (e.g., third-party servers) yet inferring on private data. However, existing HE libraries [somewhat (SWHE), leveled (LHE) or fully homomorphic (FHE)] suffer from extensive computational and memory overhead. Few performance optimized high-speed homomorphic libraries are either suffering from certain approximation issues leading to decryption errors or proven to be insecure according to recent published attacks. In this article, we propose architectural tricks to achieve performance speedup for encrypted DL inference developed with exact HE schemes without any approximation or decryption error in homomorphic computations. The main idea is to apply quantization and suitable data packing in the form of bitslicing to reduce the costly noise handling operation, Bootstrapping while achieving a functionally correct and highly parallel DL pipeline with a moderate memory footprint. Experimental evaluation on the MNIST dataset shows a significant ( $37\times$ ) speedup over the nonbitsliced versions of the same architecture. Low memory bandwidths (700 MB) of our design pipelines further highlight their promise toward scaling over larger gamut of Edge-AI analytics use cases.

The challenge of maintaining confidentiality of stored data in cloud is of utmost importance to realize the potential of cloud computing as an emerging storage solution service. Storing data in encrypted form may solve the problem, but exposes data to an adversary for each required computation. This repeated encryption decryption also diminishes the essence of cloud for storing encrypted database and huge computation power of cloud remains unused. Fully homomorphic encryption (FHE) is an effective scheme to support arbitrary operations directly on encrypted data, but has serious performance issues. In this paper, we have considered sorting on encrypted data, which is a frequently required database operation. We have investigated the feasibility of performing comparison as well as partition based sort on CPA resistant FHE data and highlight an important observation that time requirement of partition based sort on FHE data is no better than comparison based sort owing to the underlying security of the cryptosystem. We identify the recrypt operation, which is the denoising step of FHE as the main reason of costly timing requirement of such operations. We propose a FHE specific two stage sorting technique termed as Lazysort with reduced recrypt operation, which proves to be better in terms of performance on FHE data in comparison to partition as well as comparison sort. Finally, we provide some multi-core implementation results to show that with proper implementation tricks performance of FHE computations can be improved further.

Cloud provides large shared resources where users (or foundations) can enjoy the facility of storing data or executing applications. In spite of gaining convenience of large resources, storing critical data in cloud is not secured. Hence, cloud security is an important issue to make cloud useful at the enterprise level. Data encryption is a primary solution for providing confidentiality to sensitive data. However, processing of encrypted data requires extra overhead, since repeated encryption-decryption need to be performed for every simple processing on encrypted data. Hence, direct processing on encrypted cloud data is advantageous, which is supported by homomorphic encryption schemes. Fully Homomorphic Encryption (FHE) provides a method of performing arbitrary operations directly on encrypted data. This seemingly magical idea is a welcome to cloud computing. However, there are several challenges to overcome for making the technology viable in practical applications. In this paper, we make an initial effort to highlight the problem of translating algorithms that can run on unencrypted or normal data to those which operate on encrypted data. Here, we show that although FHE provides the ability to perform arbitrary computations, its complete benefit can only be obtained if they also allow to execute arbitrary algorithms on encrypted data. In this pursuit, we provide techniques to translate basic operators (like bitwise, arithmetic and relational operators), which are used for implementation of algorithms in any high level language like C. Subsequently, we address decision making and loop handling and related data structures which are vital to realize when the controlling variables are encrypted. Since, termination is a major challenge while handling encrypted data, we propose a method of handling termination by message passing between server and client.

Elliptic curve cryptography (ECC) has proven its superiority, since it was proposed in the domain of Public-Key Cryptography [1]. Further, Edwards curve adds a new paradigm to ECC in terms of speed and security against exceptional point attacks. This curve has been recently extended to Binary Edwards Curves (BEC), due to efficiency of implementation in GF(2m) fields and to harvest the advantages of a unified and complete scalar point multiplication on the family of BEC. In spite of achieving the unification, it introduces more challenges to the designer to reduce the computation time and trade-off the area in efficient way. This work reports an implementation of BEC processor with an effort to better utilize the look-up table (LUT) of the FPGA. The design further implements the ternary algorithm to increase the efficiency. However, to the best of our knowledge there exists no previous implementations of BEC on FPGA platform. The proposed design has been implemented for state-of-the-art GF(2233) fields. The performance of the design has been found to compare favorably with the existing designs on standard cell ASIC libraries, in spite of being implemented on FPGA platform.