Biblio

With the increasing awareness of privacy protection and data security, people’s concerns over the confidentiality of sensitive data still limit the application of distributed artificial intelligence. In fact, a new encryption form, called homomorphic encryption(HE), has achieved a balance between security and operability. In particular, one of the HE schemes named Paillier has been adopted to protect data privacy in distributed artificial intelligence. However, the massive computation of modular multiplication in Paillier greatly affects the speed of encryption and decryption. In this paper, we propose a fast CRT-Paillier scheme to accelerate its decryption process. We first introduce the Montgomery algorithm to the CRT-Paillier to improve the process of the modular exponentiation, and then compute the modular exponentiation in parallel by using OpenMP. The experimental results show that our proposed scheme has greatly heightened its decryption speed while preserving the same security level. Especially, when the key length is 4096-bit, its speed of decryption is about 148 times faster than CRT-Paillier.

Derivatives are key to numerous science, engineering, and machine learning applications. While existing tools generate derivatives of programs in a single language, modern parallel applications combine a set of frameworks and languages to leverage available performance and function in an evolving hardware landscape. We propose a scheme for differentiating arbitrary DAG-based parallelism that preserves scalability and efficiency, implemented into the LLVM-based Enzyme automatic differentiation framework. By integrating with a full-fledged compiler backend, Enzyme can differentiate numerous parallel frameworks and directly control code generation. Combined with its ability to differentiate any LLVM-based language, this flexibility permits Enzyme to leverage the compiler tool chain for parallel and differentiation-specitic optimizations. We differentiate nine distinct versions of the LULESH and miniBUDE applications, written in different programming languages (C++, Julia) and parallel frameworks (OpenMP, MPI, RAJA, Julia tasks, MPI.jl), demonstrating similar scalability to the original program. On benchmarks with 64 threads or nodes, we find a differentiation overhead of 3.4–6.8× on C++ and 5.4–12.5× on Julia.

The paper presents shared memory implementation of chaotic RGB LSB steganography technique, The proposed technique involves hiding the secret information into RGB components of the cover image. Chaotic logistic map has been used to generate highly random numbers for enhancing the security of embedded information. Encryption and decryption process is parallelized using OpenMP API in multicore environment, and results show significant speed up and highly scalable results even with large amount of data.

This is a feasibility study on homomorphic encryption using the TFHE library [1] in daily computing using cloud services. A basic set of arithmetic operations namely - addition, subtraction, multiplication and division were created from the logic gates provide. This research peeks into the impact of logic gates on these operations such as latency of the gates and the operation itself. Multiprocessing enhancement were done for multiplication operation using MPI and OpenMP to reduce latency.

The network attack graph is a powerful tool for analyzing network security, but the generation of a large-scale graph is non-trivial. The main challenge is from the explosion of network state space, which greatly increases time and storage costs. In this paper, three parallel algorithms are proposed to generate scalable attack graphs. An OpenMP-based programming implementation is used to test their performance. Compared with the serial algorithm, the best performance from the proposed algorithms provides a 10X speedup.

Algorithm parallelization diversifies a complicated computing task into small parts, and thus it receives wide attention when it is implemented to evolutionary algorithms (EA). This works considers a recently developed EA called the Swarm Intelligence Based (SIB) method as a benchmark to compare the performance of two types of parallel computing approaches: a CPU-based approach via OpenMP and a GPU-based approach via CUDA. The experiments are conducted to solve an optimization problem in the search of supersaturated designs via the SIB method. Unlike conventional suggestions, we show that the CPU-based OpenMP outperforms CUDA at the execution time. At the end of this paper, we provide several potential problems in GPU parallel computing towards EA and suggest to use CPU-based OpenMP for parallel computing of EA.

Hashing algorithms are used extensively in information security and digital forensics applications. This paper presents an efficient parallel algorithm hash computation. It's a modification of the SHA-1 algorithm for faster parallel implementation in applications such as the digital signature and data preservation in digital forensics. The algorithm implements recursive hash to break the chain dependencies of the standard hash function. We discuss the theoretical foundation for the work including the collision probability and the performance implications. The algorithm is implemented using the OpenMP API and experiments performed using machines with multicore processors. The results show a performance gain by more than a factor of 3 when running on the 8-core configuration of the machine.

Hashing algorithms are used extensively in information security and digital forensics applications. This paper presents an efficient parallel algorithm hash computation. It's a modification of the SHA-1 algorithm for faster parallel implementation in applications such as the digital signature and data preservation in digital forensics. The algorithm implements recursive hash to break the chain dependencies of the standard hash function. We discuss the theoretical foundation for the work including the collision probability and the performance implications. The algorithm is implemented using the OpenMP API and experiments performed using machines with multicore processors. The results show a performance gain by more than a factor of 3 when running on the 8-core configuration of the machine.