Parallel Techniques Research Articles

Introducing a Parallel Genetic Algorithm for Global Optimization Problems

The topic of efficiently finding the global minimum of multidimensional functions is widely applicable to numerous problems in the modern world. Many algorithms have been proposed to address these problems, among which genetic algorithms and their variants are particularly notable. Their popularity is due to their exceptional performance in solving optimization problems and their adaptability to various types of problems. However, genetic algorithms require significant computational resources and time, prompting the need for parallel techniques. Moving in this research direction, a new global optimization method is presented here that exploits the use of parallel computing techniques in genetic algorithms. This innovative method employs autonomous parallel computing units that periodically share the optimal solutions they discover. Increasing the number of computational threads, coupled with solution exchange techniques, can significantly reduce the number of calls to the objective function, thus saving computational power. Also, a stopping rule is proposed that takes advantage of the parallel computational environment. The proposed method was tested on a broad array of benchmark functions from the relevant literature and compared with other global optimization techniques regarding its efficiency.

Analysis and Construction of Hardware Accelerators for Calculating the Shortest Path in Real-Time Robot Route Planning

This study introduces an optimization approach for calculating the shortest path in mobile robot route planning. The proposed solution targets real-time processing requirements by offering a high-performance alternative. This is achieved by embedding in the dedicated hardware an architecture which emphasizes parallelism. Through improvements in parallel exploration techniques, our solution aims to present not only a boost in performance but also a dynamic adaptation to graph changes, accommodating randomly occurring edge insertions or deletions as environmental conditions fluctuate. We present the developed architecture alongside its results. Our method efficiently updates obstacle matrices, resulting in a remarkable 120-fold improvement for 1024-node graphs. When utilizing a cost-effective device like the Cyclone IV E, it achieves approximately 12 times the performance of software applications.

The application of the golden-angle radial sparse parallel technique in T restaging of locally advanced rectal cancer after neoadjuvant chemoradiotherapy.

To evaluate the diagnostic performance of Golden-Angle Radial Sparse Parallel (GRASP) MRI in identifying pathological stage T0-1 (ypT0-1) after neoadjuvant chemoradiotherapy (nCRT) in patients with rectal cancer, compared to T2-weighted imaging (T2WI) combined with Diffusion Weighted Imaging (DWI). In this retrospective study, 168 patients were carefully selected based on inclusion criteria that targeted individuals with biopsy-confirmed primary rectal adenocarcinoma, identified via MRI as having locally advanced disease (≥ T3 and/or positive lymph node results) prior to nCRT. Post-nCRT, all MRI images obtained after nCRT were assessed by two observers independently. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy for identifying ypT0-1 based on GRASP and T2 + DWI were calculated. Multivariable regression analysis was used to explore the factors independently associated with ypT0-1 tumor. 45 patients out of these cases were ypT0-1, and the accuracy, sensitivity, specificity, PPV, and NPV of GRASP were higher than the T2 + DWI (88% vs 74%, 93% vs 71%, 86% vs 75%, 71% vs 52% and 97% vs 88%), the AUC in identifying ypT0-1 tumor based on GRASP was 0.90 (95% CI:0.84, 0.94), which was better than the T2 + DWI (0.73; 95% CI: 0.66, 0.80). Multivariable logistic regression analysis showed that the yT stage on GRASP scans was the only factor independently associated with ypT0-1 tumor (P < 0.001). The GRASP helped distinguish ypT0-1 tumor after nCRT and can select patients who may be suitable for local excision.

Improvement of Apriori Algorithm Using Parallelization Technique on Multi-CPU and GPU Topology

Improvement of Apriori Algorithm Using Parallelization Technique on Multi-CPU and GPU Topology

Electromagnetic scattering and imaging simulation of extremely large-scale sea-ship scene based on GPU parallel technology

Aiming to solve the bottleneck problem of electromagnetic scattering simulation in the scenes of extremely large-scale seas and ships, a high-frequency method by using graphics processing unit (GPU) parallel acceleration technique is proposed. For the implementation of different electromagnetic methods of physical optics (PO), shooting and bouncing ray (SBR) and physical theory of diffraction (PTD), a parallel computing scheme based on the central processing unit (CPU)-GPU parallel computing scheme is realized to balance computing tasks. Finally, a multi-GPU framework is further proposed to solve the computational difficulty caused by the massive number of ray tubes in the ray tracing process. By using the established simulation platform, signals of ships at different seas are simulated and their images are achieved as well. It is shown that the higher sea states degrade the averaged peak signal-to-noise ratio (PSNR) of radar image.

Modeling and analysis for landing airbag–lunar soil interaction using a CPU–GPU-based FMBD-DEM computational framework

To investigate the landing performance of airbag cushion systems on the discrete lunar soil, this study proposes a novel modeling approach that couples multi-physics fields, including the flexible multibody systems (FMBS), the granular systems, and the gas effects. In this model, the GPU parallelism technique is utilized to address challenges arising from the large number of particles and complex contact detection involving large deformation and overall motions, and an efficient contact implementation between finite elements and particles is introduced. Additionally, contact modeling between airbags is realized using master–slave techniques. Then, a modified conventional serial staggered procedure is employed to couple flexible multibody dynamics (FMBD) and discrete element method (DEM) modules. Finally, a series of numerical examples is presented to validate the proposed dynamic model. The validated model successfully simulates the interaction between landing airbags and lunar soil. Subsequent parametric analysis contributes to the design optimization of the landing system. Moreover, discussions regarding computational efficiency illustrate that the proposed CPU–GPU-based FMBD-DEM is advantageous in engineering scenarios involving flexible components and large-scale granular systems.

A GPU-accelerated adaptation of the PSO algorithm for multi-objective optimization applied to artificial neural networks to predict energy consumption

Optimization research often confronts the challenge of developing time consuming processes. This article introduces an innovative approach that leverages the computational power of Graphics Processing Units (GPUs) to speed up that optimization process. We present an innovative adaptation of Particle Swarm Optimisation (PSO) to meet the requirements of multiobjective optimization problems. This approach aims to leverage the strengths of a multi-objective approach to perform energy consumption prediction using neural networks. By employing GPU parallel techniques, our method not only speeds up the optimization process but also enhances the efficiency of neural network training execution. The main advantage of our approach lies in its dual ability to simultaneously optimizing neural network architectures by determining the minimum number of hidden neurons and fitting the weights of the networks in order to achieve the lowest error. Preliminary results suggest a notable enhancement in prediction accuracy of forecasting electric energy consumption, as a result of optimizing the architecture and parameters of the neural network using the proposed method. This PSO adaptation stands out for its ability to address complex problems, increase efficiency and produce accurate predictions, making it a novel solution in Machine Learning heuristic methods for application in the solution of advanced prediction problems with time constraints from time series.

Fractal Analysis of Trabecular Alveolar Bone with Intrabony and Furcation Defects Using Periapical Dental Radiographs

Objective: Fractal analysis (FA) is a non-invasive method that quantitatively measures complex patterned geometric structures present throughout the image. Trabecular morphology of the alveolar bone and the changes occurring in the trabeculae in case of periodontitis can be detected with this method. To examine the periodontal defects in human skull bones using the FA, to compare them with healthy alveolar bone regions. Methods: Furcation and intrabony defects were artificially created in the mandible alveolar bones (n:24). Periapical X-ray images of alveolar bone regions containing teeth with defects were taken using the parallel technique. Fractal analysis was performed by box-counting method using Image J software on images from areas containing healthy and defective trabecular bone. Results: No statistically significant difference was found between the fractal values of healthy tissue and bifurcation defetcs and between the healthy tissue and intrabony defects (p&gt;0.05). Conclusion: Many factors may have affected the outcomes; patient selection, imaging methods, sample size, Region of interest (ROI) selection-location and size, individual and anatomical variations. These variables need to be standardized as much as possible and the limitations of the method need to be improved.

A Comprehensive Analysis of Parallel Genetic Algorithms

The paper introduces an optimal approach for optimizing the idle time of processors through the utilization of parallel genetic algorithms. By harnessing the power of parallelism, we propose a method that effectively minimizes idle periods in processor operations. Our approach, outlined in detail within this paper, demonstrates significant improvements in resource utilization and efficiency. Through rigorous experimentation and analysis, we illustrate the efficacy of our parallel genetic algorithm in optimizing processor idle time. The findings presented herein offer promising insights into enhancing computational resource utilization, particularly in multi-processor systems. This research contributes to the ongoing efforts in maximizing computational efficiency and lays a foundation for further advancements in parallel optimization techniques. Keywords: Optimization; Metaheuristic; Parallel Genetic algorithm; Crossover; Mutation; Selection; Evolution;

Encrypted Network Traffic Classification Using Deep and Parallel Network-In-Network Models

Network traffic classification analyzes received data packets to identify distinct application or traffic kinds. This research describes a neural network model that uses deep and parallel network-in-network (NIN) architectures to classify encrypted network data. In comparison to typical convolutional neural networks (CNN), NIN uses a micro network after each convolution layer to improve local modeling. Furthermore, NIN uses global average pooling instead of traditional fully connected layers before final classification, resulting in a considerable reduction in the number of model parameters. Our suggested solution uses deep NIN models with several MLP convolutional layers to map fixed-length packet vectors to application or traffic labels. Furthermore, a parallel decision method is created to build two sub-networks to process packet headers and packet bodies separately, taking into account that they may include different types of evidence for classification. Our investigations on the ''ISCX VPN-nonVPN'' encrypted traffic dataset demonstrate that NIN models can achieve a better balance between classification accuracy and model complexity than standard CNNs. The parallel decision technique can increase the accuracy of a single NIN model for classifying encrypted network data. Finally, the test set F1 scores of 0.983 and 0.985 are obtained for traffic characterisation and application identification, respectively

Evaluation of reduced-scale railway ballast particle size effects for different test sample sizes using monotonic triaxial tests

This study rigorously investigated the mechanical characteristics of railway ballast using largescale triaxial (LST) testing, focusing on shear strength, stress–strain relationships, and volumetric alterations by altering particle sizes under controlled conditions. Samples with a size ratio (D/dmax) of 4.7 are typically used in LST testing. However, the ballast used in Thailand typically features a maximum grain size of 70.00 mm, whereas LST testing requires samples with diameter exceeding 300 mm, which poses logistical and material challenges. To mitigate this issue, our test introduced a parallel grading method to decrease sample size while preserving essential ballast properties and morphological features. This test applied two grain-size distributions: normal grain-size distribution (NGD) and parallel grain-size distribution (PGD). The study aimed to ascertain whether the smaller-sized samples can still provide accurate assessments of mechanical behaviors. The outcomes demonstrated that both NGD and PGD approaches yielded similar stress–strain behaviors and cohesion values. A marginal difference in friction angle values of 4° was noted, suggesting the occurrence of minor variances in shear resistance, which did not significantly affect the overall mechanical properties. The findings validate the effectiveness of the parallel grading technique, thus offering practical implications for testing procedures in railway engineering.

Application of Parallel Computing on Hybrid Inundation Model, Case Study: Chiayi County Flood 23-24 August 2018

A robust 2D inundation model is needed to support flood warning systems in urban areas. The conventional 2D hydrodynamic model uses shallow water equations as the governing equation and is computationally expensive. Although the models have benefited from parallel computation techniques, some issues remain. As an alternative, many flood models have been developed using different approaches, such as Cellular Automata (CA), DEM-based (DBM), and data-driven models. The hybrid inundation model (HIM) was developed by combining the CA-DBM concepts. The purpose of this study is to implement the parallel computation technique to increase the efficiency of HIM. The model performance was evaluated using the historical flood event in Chiayi County, Taiwan. Results showed that there is no significant difference between HIM and TUFLOW in terms of flood depth estimation, even though TUFLOW included the drainage system within the analysis. These results proved that the drainage system was not working during the event. HIM and TUFLOW give underestimated flood depth prediction compared to the observed data. The main reason because the observed data was obtained from local community testimonies. Hence, there might be many uncertainties in the observed data value. Finally, the parallelization process successfully decreased the computation time of the original HIM. The computation was decreased from 450 to 11 minutes depending on the number of cores used in the simulation.

Efficient Edge Computing: A Survey of High-Throughput Concurrent Processing Strategies for Graph Data

This paper reviews the strategies for high-throughput concurrent processing of graph data in edge computing environments. As information technology rapidly advances, particularly in areas such as the Internet of Things (IoT), smart cities, and autonomous driving, the need for real-time and efficient data processing continues to grow. Edge computing is a distributed computing paradigm that can process data at or near its origin, thereby reducing network latency, improving application performance, and reducing the load on central data centers. The discussion includes the application of edge computing in graph data processing, highlighting parallel computing models such as the Bulk Synchronous Parallel (BSP) model and other parallel optimization techniques like data partitioning and task scheduling. The paper also addresses specific challenges of parallel processing in edge computing, such as resource constraints, data communication delays, and strategies for security and privacy protection. Despite the significant potential edge computing has demonstrated in processing graph data, numerous challenges remain. Future research directions involve the development of new resource optimization algorithms, low-latency communication protocols, and technologies to enhance data security, aiming to achieve more efficient and secure graph data processing. This paper provides clear directions and a foundation for the efficient implementation of graph data processing in edge computing environments, positioning edge computing to play an increasingly significant role in real-time data analysis and intelligent applications.

Multilayer Neural Network Hardware Designs for Faster Convolutional Processing

The demand for efficient hardware implementations of convolutional neural networks (CNNs) has surged with the proliferation of deep learning applications in various domains such as computer vision, natural language processing, and autonomous systems. Convolutional layers, being the fundamental building blocks of CNNs, are computationally intensive, requiring optimized hardware architectures for real-time inference tasks. This paper presents novel hardware designs targeting faster convolutional processing within multilayer neural networks. We propose a multi-faceted approach that leverages parallelism, pipelining, and hardware acceleration techniques to enhance the efficiency of convolution operations. Our design optimally exploits the inherent data-level and model-level parallelism present in CNNs to achieve high throughput while minimizing latency.

Optimization of short-term hydropower scheduling with dynamic reservoir capacity based on improved genetic algorithm and parallel computing

The optimal scheduling of hydropower plants is important for the efficient operation and management of hydropower energy systems. In the short-term optimization scheduling of river-type reservoirs, the dynamic reservoir capacity is a key factor. Currently, the short-term power generation scheduling model considering dynamic reservoir capacity was solved by the enumeration method, providing a more accurate simulation of the water flow process compared to the model considering static reservoir capacity. However, the solving process had the disadvantage of a large computational burden and excessive time consumption, which restricted application in the actual scheduling of reservoirs. Therefore, to solve the time-consuming problem of repeated calculation of hydrodynamics model in optimization scheduling, an improved genetic algorithm including adaptive and elitist-selection strategies combined with the parallel computing technique was proposed. The optimal scheduling plan generated by this model were compared with the optimization results of the enumeration method and the genetic algorithm scheduling model. The results showed that the optimal power generation selected by the three models were extremely similar and the optimization model could find the optimal value for the scheduling period. In terms of computing speed, the improved genetic algorithm parallel technique took approximately 0.8 h, around 246 times faster than the enumeration method and 20 times faster than the model without using parallel computing. It demonstrates that the present algorithm is practically usable, significantly enhancing the optimization efficiency of scheduling solutions. Furthermore, it satisfies the engineering application efficiency requirements and can provide technical support for the subsequent power generation scheduling of the Three Gorges Reservoir.

PnV: An Efficient Parallel Consensus Protocol Integrating Proof and Voting

Consensus protocols, as crucial components of blockchain technology, play a vital role in ensuring data consistency among distributed nodes. However, the existing voting-based and proof-based consensus protocols encounter scalability issues within the blockchain system. Moreover, most consensus protocols are serialized, which further limits their scalability potential. To address this limitation, parallelization methods have been employed in both types of consensus protocols. Surprisingly, however, novel fusion consensus protocols demonstrate superior scalability compared with these two types but lack the utilization of parallelization techniques. In this paper, we present PnV, an efficient parallel fusion protocol integrating proof-based and voting-based consensus features. It enhances the data structure, consensus process, transaction allocation, and timeout handling mechanisms to enable concurrent block generation by multiple nodes within a consensus round. Experimental results demonstrate that PnV exhibits superior efficiency, excellent scalability, and acceptable delay compared with Proof of Vote (PoV) and BFT-SMART. Moreover, at the system level, the performance of the PnV-based blockchain system optimally surpasses that of the FISCO BCOS platform. Our proposed protocol contributes to advancing blockchain technology by providing a more efficient and practical solution for achieving decentralized consensus in distributed systems.

Application of parallel artificial membrane permeability assay technique and chemometric modeling for blood-brain barrier permeability prediction of protein kinase inhibitors.

Aim: This study aims to investigate the passive diffusion of protein kinase inhibitors through the blood-brain barrier (BBB) and to develop a model for their permeability prediction. Materials & methods: We used the parallel artificial membrane permeability assay to obtain logPe values of each of 34 compounds and calculated descriptors for these structures to perform quantitative structure-property relationship modeling, creating different regression models. Results: The logPe values have been calculated for all 34 compounds. Support vector machine regression was considered the most reliable, and CATS2D_09_DA, CATS2D_04_AA, B04[N-S]and F07[C-N] descriptors were identified as the most influential to passive BBB permeability. Conclusion: The quantitative structure-property relationship-support vector machine regression model that has been generated can serve as an efficient method for preliminary screening of BBB permeability of new analogs.

Strong–strong beam–beam simulations with lattices of circular e+e- colliders

Strong–strong beam–beam simulations for circular colliders are very time-consuming. Thus far, such simulations have predominantly utilized a linear transfer matrix to represent the full nonlinear lattice. Otherwise, when nonlinear lattices were included, beam–beam interactions were usually studied by using weak–strong or quasi-strong–strong numerical models. The next-generation circular colliders, including FCC-ee at CERN, CEPC in China, and the Super Tau-Charm Factories in China and Russia, will leverage the crab waist collision scheme with a large Piwinski angle. This scheme has already been employed in the two currently operational colliders, DAFNE and SuperKEKB. Accurate predictions of the performance of such colliders call for strong–strong beam–beam simulations, taking into account the interplay of the beam–beam interaction with lattice nonlinearities, impedances, and other collective effects. To answer this call, a novel code APES-T that allows integrating the direct strong–strong beam–beam simulations with element-by-element nonlinear tracking has been developed. To overcome the limitations in computing time, APES-T employs parallelization techniques. Specifically, the beam–beam module is parallelized using MPI programming, while the element-by-element tracking modules are parallelized using GPU programming. In this paper, we present the development of the APES-T code, its benchmark against the SAD code, and its application to the CEPC in investigating the interplay between beam–beam interaction, beamstrahlung, and lattices.

Energy efficiency and performance analysis of a legacy atomic scale materials modeling simulator (VASP)

This work tackles the performance and energy consumption analysis of a legacy scientific application, the VASP (Vienna Ab-initio Simulation Package), an application commonly used by physicists and chemists for modeling materials at the atomic scale. Many of these scientific applications have been implemented in Fortran, where energy metrics instrumentation is not straightforward. We obtained performance figures (execution time and energy consumption) by instrumenting the source code using EML. This energy measurement library has been modified to introduce Fortran interfaces for these metrics. The analysis was carried out using different matrix algebra libraries, parallelization techniques, and hardware platforms, emphasizing on the MPI, OpenMP, and CUDA parallel implementations of the algorithms used in VASP. We employ various material specifications (atomic structures) and molecular sizes of a silicon-based crystal to create a set of benchmarks for these specifications, leading to some recommendations for final users regarding performance improvements. The proposed benchmarking technique assists the user in selecting the right combination of problem size, compilers, and parallelization options available in VASP. For a given system platform, the user will be able to determine not only the architecture to use (GPU or multicore processors), but also the appropriate library and parallelization according to the atomic structure and molecular size.

Optimizing Neural Network Inference in Edge Robotics by Harnessing FPGA Hardware Acceleration

This research explores advancements in neural network inference for edge robotics through the implementation of hardware acceleration using Field-Programmable Gate Arrays (FPGA). Leveraging FPGA's versatility in custom hardware design, the research optimizes performance and energy efficiency by utilizing features such as Digital Signal Processors (DSP), Multiply-Accumulate (MAC) arrays, parallel processing units, Convolutional Neural Networks (CNN), and custom accelerators. The computational power of FPGA-based architectures is harnessed to significantly enhance the speed and efficiency of neural network inference tasks in edge robotics applications. Key aspects include the utilization of DSP blocks for efficient signal processing, MAC arrays for high-speed matrix operations crucial in neural network computations, and parallel processing units for concurrent execution of multiple tasks, improving overall system throughput. The integration of CNN directly onto FPGA hardware allows for minimal-latency inference, making it ideal for real-time applications in edge robotics scenarios. Custom accelerators tailored to specific neural network operations further optimize performance and resource utilization. The research involves the meticulous design and synthesis of hardware modules tailored to the target FPGA platform. Through rigorous optimization and parallelization techniques, the goal is to maximize resource utilization while minimizing latency and power consumption. The experimental findings showcase remarkable results, including a substantial increase in throughput, with values reaching up to 150 images/sec, and impressive accuracy rates exceeding 95% in classification tasks.