Multilevel Pipeline Research Articles

Compact and Low-Latency FPGA-Based Number Theoretic Transform Architecture for CRYSTALS Kyber Postquantum Cryptography Scheme

In the modern era of the Internet of Things (IoT), especially with the rapid development of quantum computers, the implementation of postquantum cryptography algorithms in numerous terminals allows them to defend against potential future quantum attack threats. Lattice-based cryptography can withstand quantum computing attacks, making it a viable substitute for the currently prevalent classical public-key cryptography technique. However, the algorithm’s significant time complexity places a substantial computational burden on the already resource-limited chip in the IoT terminal. In lattice-based cryptography algorithms, the polynomial multiplication on the finite field is well known as the most time-consuming process. Therefore, investigations into efficient methods for calculating polynomial multiplication are essential for adopting these quantum-resistant lattice-based algorithms on a low-profile IoT terminal. Number theoretic transform (NTT), a variant of fast Fourier transform (FFT), is a technique widely employed to accelerate polynomial multiplication on the finite field to achieve a subquadratic time complexity. This study presents an efficient FPGA-based implementation of number theoretic transform for the CRYSTAL Kyber, a lattice-based public-key cryptography algorithm. Our hybrid design, which supports both forward and inverse NTT, is able run at high frequencies up to 417 MHz on a low-profile Artix7-XC7A100T and achieve a low latency of 1.10μs while achieving state-of-the-art hardware efficiency, consuming only 541-LUTs, 680 FFs, and four 18 Kb BRAMs. This is made possible thanks to the newly proposed multilevel pipeline butterfly unit architecture in combination with employing an effective coefficient accessing pattern.

A review on hazards and risks to pipeline operation under transporting hydrogen energy and hydrogen-mixed natural gas

As an efficient and clean fuel, hydrogen energy plays an important role in relieving the energy crisis and achieving the orientation of zero carbon emissions. Transportation is the key link in the construction of hydrogen energy infrastructure. For large-scale and long-distance transportation of hydrogen, pipeline transportation has the advantages of high efficiency and cost saving. While using the existing natural gas pipeline to transport hydrogen, it would economize the economic cost, time cost and labor cost. However, the transportation of hydrogen may bring more hazards and risks. Based on the investigation of a large number of literatures, the research advance in hydrogen embrittlement, leakage, combustion and explosion risk of hydrogen and hydrogen-mixed natural gas pipelines was reviewed. The mechanism, research means and evaluation methods of hydrogen embrittlement, as well as the experimental and numerical simulation research results of leakage, combustion and explosion were discussed in detail. The definite and important conclusions include: (1) For buried hydrogen-mixed natural gas transportation pipeline, the leakage rate of hydrogen and methane is the same, the formation of the leakage crater is foreign to the nature of leakage gas. (2) When adding less than 25 volume percentage of hydrogen into the natural gas pipelines, the explosion risk would not be increased. Future research should focus on the risk prediction, quantitative risk assessment, intelligent monitoring, and explosion-suppression technical measures of hydrogen and hydrogen-mixed natural gas transportation pipelines, so as to establish comprehensive and multi-level pipeline safety protection barriers.

Scalable and Efficient Full-Graph GNN Training for Large Graphs

Graph Neural Networks (GNNs) have emerged as powerful tools to capture structural information from graph-structured data, achieving state-of-the-art performance on applications such as recommendation, knowledge graph, and search. Graphs in these domains typically contain hundreds of millions of nodes and billions of edges. However, previous GNN systems demonstrate poor scalability because large and interleaved computation dependencies in GNN training cause significant overhead in current parallelization methods. We present G3, a distributed system that can efficiently train GNNs over billion-edge graphs at scale. G3 introduces GNN hybrid parallelism which synthesizes three dimensions of parallelism to scale out GNN training by sharing intermediate results peer-to-peer in fine granularity, eliminating layer-wise barriers for global collective communication or neighbor replications as seen in prior works. G3 leverages locality-aware iterative partitioning and multi-level pipeline scheduling to exploit acceleration opportunities by distributing balanced workload among workers and overlapping computation with communication in both inter-layer and intra-layer training processes. We show via a prototype implementation and comprehensive experiments that G3 can achieve as much as 2.24x speedup in a 16-node cluster, and better final accuracy over prior works.

A multilevel screening pipeline in zebrafish identifies therapeutic drugs for GAN

Giant axonal neuropathy (GAN) is a fatal neurodegenerative disorder for which there is currently no treatment. Affecting the nervous system, GAN starts in infancy with motor deficits that rapidly evolve toward total loss of ambulation. Using the gan zebrafish model that reproduces the loss of motility as seen in patients, we conducted the first pharmacological screening for the GAN pathology. Here, we established a multilevel pipeline to identify small molecules restoring both the physiological and the cellular deficits in GAN. We combined behavioral, in silico, and high‐content imaging analyses to refine our Hits to five drugs restoring locomotion, axonal outgrowth, and stabilizing neuromuscular junctions in the gan zebrafish. The postsynaptic nature of the drug's cellular targets provides direct evidence for the pivotal role the neuromuscular junction holds in the restoration of motility. Our results identify the first drug candidates that can now be integrated in a repositioning approach to fasten therapy for the GAN disease. Moreover, we anticipate both our methodological development and the identified hits to be of benefit to other neuromuscular diseases.

Fast and Accurate Detection of COVID-19 Along With 14 Other Chest Pathologies Using a Multi-Level Classification: Algorithm Development and Validation Study.

BackgroundCOVID-19 has spread very rapidly, and it is important to build a system that can detect it in order to help an overwhelmed health care system. Many research studies on chest diseases rely on the strengths of deep learning techniques. Although some of these studies used state-of-the-art techniques and were able to deliver promising results, these techniques are not very useful if they can detect only one type of disease without detecting the others.ObjectiveThe main objective of this study was to achieve a fast and more accurate diagnosis of COVID-19. This study proposes a diagnostic technique that classifies COVID-19 x-ray images from normal x-ray images and those specific to 14 other chest diseases.MethodsIn this paper, we propose a novel, multilevel pipeline, based on deep learning models, to detect COVID-19 along with other chest diseases based on x-ray images. This pipeline reduces the burden of a single network to classify a large number of classes. The deep learning models used in this study were pretrained on the ImageNet dataset, and transfer learning was used for fast training. The lungs and heart were segmented from the whole x-ray images and passed onto the first classifier that checks whether the x-ray is normal, COVID-19 affected, or characteristic of another chest disease. If it is neither a COVID-19 x-ray image nor a normal one, then the second classifier comes into action and classifies the image as one of the other 14 diseases.ResultsWe show how our model uses state-of-the-art deep neural networks to achieve classification accuracy for COVID-19 along with 14 other chest diseases and normal cases based on x-ray images, which is competitive with currently used state-of-the-art models. Due to the lack of data in some classes such as COVID-19, we applied 10-fold cross-validation through the ResNet50 model. Our classification technique thus achieved an average training accuracy of 96.04% and test accuracy of 92.52% for the first level of classification (ie, 3 classes). For the second level of classification (ie, 14 classes), our technique achieved a maximum training accuracy of 88.52% and test accuracy of 66.634% by using ResNet50. We also found that when all the 16 classes were classified at once, the overall accuracy for COVID-19 detection decreased, which in the case of ResNet50 was 88.92% for training data and 71.905% for test data.ConclusionsOur proposed pipeline can detect COVID-19 with a higher accuracy along with detecting 14 other chest diseases based on x-ray images. This is achieved by dividing the classification task into multiple steps rather than classifying them collectively.

Sharing of Genes and Pathways Across Complex Phenotypes: A Multilevel Genome-Wide Analysis.

Evidence from genome-wide association studies (GWAS) suggest that pleiotropic effects on human complex phenotypes are very common. Recently, an atlas of genetic correlations among complex phenotypes has broadened our understanding of human diseases and traits. Here, we examine genetic overlap, from a gene-centric perspective, among the same 24 phenotypes previously investigated for genetic correlations. After adopting the multilevel pipeline (freely available at http://grass.cgs.hku.hk/limx/kgg/), which includes intragenic single nucleotide polymorphisms (SNPs), genes, and gene-sets, to estimate genetic similarities across phenotypes, a large amount of sharing of several biologically related phenotypes was confirmed. In addition, significant genetic overlaps were also found among phenotype pairs that were previously unidentified by SNP-level approaches. All these pairs with new genetic links are supported by earlier epidemiological evidence, although only a few of them have pleiotropic genes in the GWAS Catalog. Hence, our gene and gene-set analyses are able to provide new insights into cross-phenotype connections. The investigation on genetic sharing at three different levels presents a complementary picture of how common DNA sequence variations contribute to disease comorbidities and trait manifestations.

Investigation of the hydrodynamic processes of a centrifugal pump in a geothermal system

The hydrodynamic and thermodynamic processes of geothermal well extraction are investigated and presented in this paper. The paper presents mathematical models for a multi-level centrifugal pump and pipeline system. The mathematical models were used to evaluate gas (nitrogen) emission in water and its effects on hydrodynamic processes. Experimental studies and mathematical modelling showed that the gas content of the fluid increases the pressure and flow pulsations within a centrifugal pump. The variation in the height of the liquid column in extraction has an influence on characteristics of the multistage centrifugal pump used in wells.

An architecture of entropy decoder, inverse quantiser and predictor for multi-standard video decoding

A VLSI architecture for entropy decoder, inverse quantiser and predictor is proposed in this article. This architecture is used for decoding video streams of three standards on a single chip, i.e. H.264/AVC, AVS (China National Audio Video coding Standard) and MPEG2. The proposed scheme is called MPMP (Macro-block-Parallel based Multilevel Pipeline), which is intended to improve the decoding performance to satisfy the real-time requirements while maintaining a reasonable area and power consumption. Several techniques, such as slice level pipeline, MB (Macro-Block) level pipeline, MB level parallel, etc., are adopted. Input and output buffers for the inverse quantiser and predictor are shared by the decoding engines for H.264, AVS and MPEG2, therefore effectively reducing the implementation overhead. Simulation shows that decoding process consumes 512, 435 and 438 clock cycles per MB in H.264, AVS and MPEG2, respectively. Owing to the proposed techniques, the video decoder can support H.264 HP (High Profile) 1920 × 1088@30fps (frame per second) streams, AVS JP (Jizhun Profile) 1920 × 1088@41fps streams and MPEG2 MP (Main Profile) 1920 × 1088@39fps streams when exploiting a 200 MHz working frequency.

Dynamic programming implementation on array processor architectures

Dynamic Programming (DP) applies to many signal and image processing applications including boundary following, the Viterbi algorithm, dynamic time warping, etc. This paper presents an array processor implementation of generic dynamic programming. Our architecture is a SIMD array attached to a host computer. The processing element of the architecture is based on an ASIC design opting for maximum speed-up. By adopting a torus interconnection network, a dual buffer structure, and a multilevel pipeline, the performance of the DP chip is expected to reach the order of several GOPS. The paper discusses both the dedicated hardware design and the data flow control of the DP chip and the total array.

A problem-oriented computer system for ionogram reduction

It is the aim of work at the institute to develop methods for fully automatic reduction of modern ionograms obtained in real time but also for interactive reduction of photographic ionograms in the archives. Tasks which must be resolved in this program are listed-up. Some algorithms and software parts are described. Processing of information and software needed for establishing stations are discussed. Structures, hardware and software needed for resolving these tasks are indicated. For the particular requirements of real time reduction a ‘multilevel pipeline computer system’ is proposed. Adaptation and learning for adapting a computer system to the type of input data are discussed.