Parallel Processing Tasks Research Articles

Enhanced photonic reservoir computing using an optically injected VCSEL with random polarized optical feedback.

We propose and numerically demonstrate a photonic time-delay reservoir computing (TDRC) system exhibiting enhanced parallel task processing performance, where an optically injected vertical-cavity surface-emitting laser (VCSEL) under random distributed optical feedback acts as the reservoir computer. To assess its effectiveness, we perform two benchmark tasks including chaotic time-series prediction and waveform recognition task, where the TDRC is associated with two different random feedback structures, i.e., orthogonally polarized optical feedback (OPOF) and parallelly polarized optical feedback (PPOF). Benefiting from the enhanced nonlinearity offered by the random distributed optical feedback, the proposed TDRC excels at parallel task processing with the PPOF structure, whereas the performance of the OPOF structure may be deteriorated. Additionally, we reveal the effect of the injection strength, feedback strength, pump current, and number of virtual nodes on the proposed TDRC. Our work paves the way for the performance enhancement of parallel task processing based on polarization multiplexing in a VCSEL-based TDRC.

Photonic reservoir computing for parallel task processing based on a feedback-free spin-polarized VCSEL

Time delayed reservoir computing (RC) is a novel artificial neural network that is easy to implement in hardware due to its extremely simple structure. Because of its time-division multiplexed information processing, laser-based photonic time-delayed RCs usually realize parallel processing with polarization/wavelength multiplexing. However, the performance of two different tasks is difficult to regulate separately and simultaneously in the time delayed RC system, especially for the chip-scale configuration. Here, we propose a feedback-free RC system based on a spin-polarized vertical-cavity surface-emitting semiconductor laser (VCSEL), which simplifies the whole system structure and can process time series prediction and waveform recognition tasks in parallel, with employing the input and output coding to provide the effect from past states. By separately setting the number of past states introduced by the coding for the two tasks, the performance of the two tasks can be adjusted respectively. Furthermore, by appropriately tuning the pump polarization ellipticity which is the unique feature for the spin-polarized VCSEL, the computational ability of the proposed RC can be focused on one of the two parallel tasks. Therefore, the proposed RC system is capable of dealing with different tasks with high performance, and also expected to provide a viable solution for integrated neuromorphic computing systems due to its compact, feedback-free structure.

Application of support vector machine system introducing multiple submodels in data mining

As the information age develops, the scale and form of data become more diverse and diverse. Therefore, people need to use effective means to process information. For large-scale data mining problems, a clustering-based kernel matrix inner product filtering method is introduced to decompose the original quadratic programming problem into multiple sub-problems to support parallel training. And a Spark-based multiple submodels parallel support vector machine is proposed. By introducing open-source tools such as OpenCV, image feature extraction can be performed on large-scale video data. Finally, combined with the designed parallel support vector machine algorithm, video facial and expression recognition is carried out. These experiments confirmed that the research method achieved a maximum acceleration ratio of 2090 times when processing Covtype datasets. The research model could achieve an accuracy of over 99 %. Under the maximum data scale experiment, the research model improved prediction accuracy by 21 percentage points with an acceptable additional time cost of about 4 min only. Task parallel processing could more fully utilize cluster performance, increasing by approximately 3.5 m/s2 from 30 to 150 cores. The research model had the highest recognition accuracy for facial expressions, further demonstrating the effectiveness and superiority of this method. The research method has improved the efficiency of big data analysis and mining, and is of great significance in parallel analysis of video data.

Abstract WMP25: Implementing a Rapid Response Model to Improve Efficiency of In-Hospital Stroke Code Processes

Introduction: In-hospital stroke (IHS), which occurs during hospitalization for other conditions, presents unique challenges and complexities. Studies have shown IHS is associated with worse patient outcomes, e.g., higher morbidity and mortality, compared to community-onset stroke treated in the emergency department. Time from initial symptom recognition to diagnostic imaging is critical for implementing first line therapy. During IHS, efficient and consistent performance of time-sensitive treatments relies on well-defined roles and parallel processing of tasks. Method: A QI initiative aimed at reducing time from initial stroke symptom recognition to initiation of a non-contrast CT (NCCT) scan was implemented at the start of FY 2022 at our CSC. Focus included broad education of all hospital personnel and their defined role in IHS recognition and care. Program analysis included retrospective and prospective data collected from patients’ electronic health records (EHR) and daily EHR audits for comparison of pre- and post-implementation data (FY 2021 and FY 2022, respectively). Data collection included time of stroke code call (symptom recognition), time of NCCT, and compliance of turn-around-time (TAT) between data points, with a goal of < 20 minutes. Results: An overall increase in the number of IHS code calls between FY 2021 and FY 2022 (n=82 vs. n=106, respectively) was appreciated, with an overall decrease in mean time from stroke recognition to NCCT ( M =22 vs. M =18.3 minutes, respectively). Overall compliance of TAT between these two data points increased from 53% to 73% post QI implementation. Additionally, we experienced an improvement in IHS activation time to IV thrombolytics administration time with a median time of 64.5 minutes (FY 2021, n=8) to median time of 59 minutes (FY 2022). Conclusion: Implementing a system wide QI initiative for IHS with well-defined roles and expectations, and parallel processing of tasks can lead to early recognition and activation of IHS code calls. Improved symptom recognition, code activation, and coordinated response efforts improved time and compliance to initial imaging as well as time to thrombolytic administration in our patient population.

Bidynamical all-optical reservoir computing for parallel task processing.

A bidynamical all-optical reservoir computing (RC) system for parallel task processing is proposed based on a unidirectional semiconductor optical amplifier optical fiber loop. The polarization dynamics and intensity dynamics are excited by the input signals injected into the reservoir via phase modulation and intensity modulation, respectively. Simultaneous computation of two independent tasks is implemented based on the dynamical responses in polarization and intensity of the optical fiber loop. To our knowledge, this is the first time that two kinds of dynamical responses of an all-optical RC system are used as independent task processing channels to implement parallel task processing. The proposed RC system can achieve good parallel task processing performance with low system cost.

A novel multimodal multi-objective optimization algorithm for multi-robot task allocation

Multi-robot task allocation (MRTA) is widely used in various fields and plays an important role in some complex task environments due to its ability to distribute parallel processing tasks. However, the multi-robot cooperative system is susceptible to actual environments or preferences of decision-makers. Therefore, providing enough solutions/schemes in the MRTA is important. To improve the reliability and feasibility of obtained solution set, an improved multimodal multi-objective differential evolution algorithm hybrid with a simulated annealing algorithm (IMMODE-SA) is proposed to solve MRTA problems in this study. In the proposed IMMODE-SA, a novel population initialization method is used to improve the population quality, and a redundant solution deletion method is employed to delete redundant solutions during the search process. Moreover, a simulated annealing algorithm is utilized to improve the exploitation capability in the last generation of evolutionary process. To verify the performance of the proposed algorithm, extensive simulation experiments are conducted on three MRTA instances. Experimental results show that the proposed algorithm performs better than other competitors on MRTA instances in terms of Hypervolume (HV). Also, the validity of the proposed algorithm is demonstrated via three experiments and experimental analysis results indicate that the IMMODE-SA can provide more equivalent optimal schemes to decision makers. Finally, it is crucial to solve MRTA problems with time window constraints.

A Novel Resource Scheduler for Resource Allocation and Scheduling in Big Data Using Hybrid Optimization Algorithm at Cloud Environment

Big Medical Data (BMD) is generated by cellular telephones, clinics, academics, suppliers, and organizations. Collecting, finding, analyzing, and managing the big data to make people's lives better, comprehending novel illnesses, and treatments, predicting results at initial phases, and making real-time choices are the actual issues in healthcare systems. Dealing with big medical data in resource scheduling is a major issue that aims to offer higher quality healthcare services. Hadoop MapReduce has been widely used for parallel processing of large data tasks and efficient job scheduling. The number of big data tasks is constantly growing; it is becoming more essential to minimize their energy usage to reduce the environmental effect and operating expenses. Hence to overcome these disadvantages, we propose a novel resource scheduler for big data using a Hybrid 2-GW Optimization Algorithm (H2-GWOA). We employ the Improved GlowWorm Swarm Optimization Algorithm (IGSOA) and Mean GreyWolf Optimization Algorithm (MGWOA) for optimizing the MapReduce framework in heterogeneous big data. The CloudSim platform was used for the simulations. The performance of the proposed scheduler is proved to be better than the conventional methods in terms of metrics like latency, makespan, resource utilization, skewness, and Central Processing Unit (CPU) consumption.

A New Integrated Analysis Suite for Fast-Ion Study in KSTAR

An integrated workflow for fast-ion analysis was developed by adapting the One Modeling Framework for Integrated Task (OMFIT) workflow manager to support a standard and unified analysis platform for KSTAR users. The newly established analysis suite offers a graphical user interface–based workflow to enable users to readily access and handle experimental data archived in various data formats and servers. Further, users can analyze the data by importing modules designed for conducting certain tasks, such as profile fitting, equilibrium reconstruction, and postprocessing of tokamak data. The procedures for preparing the inputs for fast-ion simulations are streamlined by a common workflow manager, which enables the parallel processing of various tasks to efficiently analyze large fast-ion datasets. The OMFIT platform comprises a flexible Python-based application that enables users to freely manipulate the Python scripts for applications that are unavailable in the standard workflow. The framework also offers mapping tools to translate the output data into the Integrated Modeling and Analysis Suite format to maintain application compatibility for future ITER burning plasma experiments.

Efficient optical reservoir computing for parallel data processing

We propose and experimentally demonstrate an optical reservoir computing system in free space, using second-harmonic generation for nonlinear kernel functions and a scattering medium to enhance reservoir nodes interconnection. We test it for one-step and multi-step predication of Mackey-Glass time series with different input-mapping methods on a spatial light modulator. For one-step prediction, we achieve 1.8 × 10-3 normalized mean squared error (NMSE). For the multi-step prediction, we explore two different mapping methods: linear-combination and concatenation, achieving 16-step prediction with NMSE as low as 3.5 × 10-4. Robust and superior for multi-step prediction, our approach and design have potential for parallel data processing tasks such as video prediction, speech translation, and so on.

Key Technologies of Media Big Data in-Depth Analysis System Based on 5G Platform

To meet the needs of large-scale users for personalized streaming media services with high speed, low delay, and high quality in a 5G mobile network environment, this paper studies the resource allocation mechanism of streaming media based on a 5G network from the perspective of user demand prediction, which can alleviate the pressure of mobile network, improve the utilization rate of streaming media resources and the quality of user service experience. The augmented reality visualization of large-scale social media data must rely on the computing power of distributed clusters. This paper constructs a distributed parallel processing framework in a high-performance cluster environment, which adopts a loosely coupled organizational structure. Each module can be combined, called, and expanded arbitrarily under the condition of following a unified interface. In this paper, the algebraic method of parallel computing algorithm is innovatively proposed to describe parallel processing tasks and organize and call large-scale data-parallel processing operators, which effectively supports the business requirements of large-scale parallel processing of large-scale spatial social media data and solves the bottleneck of large-scale spatial social media data-parallel processing.

An Improved Strategy for Task Scheduling in the Parallel Computational Alignment of Multiple Sequences.

Task scheduling in parallel multiple sequence alignment (MSA) through improved dynamic programming optimization speeds up alignment processing. The increased importance of multiple matching sequences also needs the utilization of parallel processor systems. This dynamic algorithm proposes improved task scheduling in case of parallel MSA. Specifically, the alignment of several tertiary structured proteins is computationally complex than simple word-based MSA. Parallel task processing is computationally more efficient for protein-structured based superposition. The basic condition for the application of dynamic programming is also fulfilled, because the task scheduling problem has multiple possible solutions or options. Search space reduction for speedy processing of this algorithm is carried out through greedy strategy. Performance in terms of better results is ensured through computationally expensive recursive and iterative greedy approaches. Any optimal scheduling schemes show better performance in heterogeneous resources using CPU or GPU.

Design and implementation of reconfigurable ASK and FSK modulation and demodulation algorithm on FPGA (Field Programmable Gate Array)

Our modern world is completely driven by our ability to communicate all over the world with great precision, accuracy, and the least amount of time possible. Communication systems are used vastly in our worlds such as radar, aerospace, naval/maritime communication, underwater communication, mobile communication, or even in outer space such as satellite communication or space missions. Different communication system needs different types of modulation techniques. Design and implementation of reconfigurable modulators on cyclone-II FPGA (Field-programmable gate array) are proposed in this paper, wherein the type of modulations and demodulation can be dynamically reconfigured on the fly based on the requirement at any particular instance. The demodulator is intelligent enough to demodulate any modulated signal. FPGA or Field Programmable Gate Array is a computing device that can be programmed like CPU (Central Processing Unit) but fasted than CPU in parallel processing and DSP (Digital Signal Processing) related tasks. Consisting method of FPGA-based modulation does not support multiple modulation techniques at the same time and is not fast as they used Simulink-based simulation. This study has proposed FPGA based alternative to that, which can embody all the modulation techniques at once, making it way more versatile, cost-effective, and easy to use and test. The process begins with designing an algorithm. Then this algorithm has to be simulated. The algorithm is designed for both modulation and demodulation. In the modulation part, it is just simple logic to create different phrases and different frequencies of sinusoidal waves. But in demodulation, based on the frequency it can detect ASK (Amplitude Shift Keying) or FSK (Frequency Shift Keying) signal all by itself. Depending on the period of the signal, the algorithm can decide whether it is an ASK or FSK signal. After deciding it is an ASK of FSK signal, it starts to demodulate the signal. The outcome is, this FPGA-based modulator and demodulator are cheap, easy to configure, and can demodulate any type of demodulated signal. As it is reconfigurable, it's easy to deploy in any situation.

Comparative analysis of software optimization methods in context of branch predication on GPUs

General Purpose computing for Graphical Processing Units (GPGPU) technology is a powerful tool for offloading parallel data processing tasks to Graphical Processing Units (GPUs). This technology finds its use in variety of domains – from science and commerce to hobbyists. GPU-run general-purpose programs will inevitably run into performance issues stemming from code branch predication. Code predication is a GPU feature that makes both conditional branches execute, masking the results of incorrect branch. This leads to considerable performance losses for GPU programs that have large amounts of code hidden away behind conditional operators. This paper focuses on the analysis of existing approaches to improving software performance in the context of relieving the aforementioned performance loss. Description of said approaches is provided, along with their upsides, downsides and extents of their applicability and whether they address the outlined problem. Covered approaches include: optimizing compilers, JIT-compilation, branch predictor, speculative execution, adaptive optimization, run-time algorithm specialization, profile-guided optimization. It is shown that the aforementioned methods are mostly catered to CPU-specific issues and are generally not applicable, as far as branch-predication performance loss is concerned. Lastly, we outline the need for a separate performance improving approach, addressing specifics of branch predication and GPGPU workflow.

Control of Fork-Join Processing Networks with Multiple Job Types and Parallel Shared Resources

A fork-join processing network is a queueing network in which tasks associated with a job can be processed simultaneously. Fork-join processing networks are prevalent in computer systems, healthcare, manufacturing, project management, justice systems, and so on. Unlike the conventional queueing networks, fork-join processing networks have synchronization constraints that arise because of the parallel processing of tasks and can cause significant job delays. We study scheduling in fork-join processing networks with multiple job types and parallel shared resources. Jobs arriving in the system fork into arbitrary number of tasks, then those tasks are processed in parallel, and then they join and leave the network. There are shared resources processing multiple job types. We study the scheduling problem for those shared resources (i.e., which type of job to prioritize at any given time) and propose an asymptotically optimal scheduling policy in diffusion scale.

Electrophysiological evidence against parallel motor processing during multitasking.

We combined behavioral measures with electrophysiological measures of motor activation (i.e., lateralized readiness potentials, LRPs) to disentangle the relative contribution of premotor and motor processes to multitasking interference in the prioritized processing paradigm. Specifically, we presented stimuli of two tasks (primary and background task) in each trial, but participants were instructed to perform the background task only if the primary task required no response. As expected, task performance was substantially influenced by a task probability manipulation: Background task responses were faster, psychological refractory period effects were smaller, and interference from the second task (i.e., backward compatibility effects) was larger when there was a larger probability that this task required a response. Critically, stimulus-locked and response-locked LRP analyses indicate that these behavioral effects of parallel processing were not driven by background task motor processing (e.g., motoric response activation) taking place during primary task processing. Instead, the LRP results suggest that these effects were exclusively localized during premotor stages of processing (e.g., response selection). Thus, the present results generally provide evidence for multitasking accounts allowing parallel task processing during response selection, whereas the task-specific motor responses are activated in a serial manner. One plausible account is that multiple task information sources can be processed in parallel, with sharing of limited cognitive resources depending on task relevance, but a primary and still active task goal prevents motor activation related to the goals of other tasks in order to avoid outcome conflict.

An Extensionally Equivalence-ensured Language for Task Parallel Processing with Backtracking-based Load Balancing

An Extensionally Equivalence-ensured Language for Task Parallel Processing with Backtracking-based Load Balancing

Investigation on Spectral Indices and Soft Classifiers-Based Water Body Segmentation Approaches for Satellite Image Analysis

The emerging threat for eco-sustainability has led to a breakthrough in satellite image analyses and such instantaneous monitoring of hazards could replenish the rejuvenation of natural ecosystem. Generally, the satellite images are huge dimensional data with numerous bands of specific details about the observed region. To apply immediate precautionary measures for environmental hazards and natural devastations, deploying a cloud-based intelligent web service for handling real time satellite image processing is inevitable. Therefore, the cloud implementation could afford integrated huge storage and parallel data processing tasks, the outcome of instantaneous satellite image processing relies with the effective data processing methods of less complexity. In this regard to address a major hazard of today which is drought monitoring, this paper focuses on developing an effective water segmentation method for such geospatial cloud web services. The Landsat 8 images of Sambhar lake region has been chosen for exploiting the water segmentation results. Most prevalent approaches from Spectral indices and unsupervised clustering such as normalized difference water index (NDWI), modified normalized difference water index, fuzzy C means, K-Means, Adaptive Regularized kernel fuzzy C means (ARKFCM) and simple linear iterative clustering-based superpixel segmentation (SLIC-SUPER) are compared, respectively. On comparative assessment using standard image quality assessment metrics, NDWI and ARKFCM outstands the rest with more accurate water body delineation. However, based on reduced computational complexity and instant localization, NDWI of spectral indexing approach clearly portray the significance of spectral influence in water body segmentation from satellite images. And it can be adapted as a persistent choice for instantaneous water body segmentation in a cloud-centered geospatial module.

Improving the performance of processing special computing tasks using an asynchronous actor model

The use of third-party ready-made solutions to perform special computing tasks (SCT): processing text arrays in a natural language, the use of cryptography methods in data transmission networks, machine learning in solving applied problems, scientific computations and many other ones is a common practice in modern development of software systems. However, the use of third-party ready-made software solutions often leads to their poor optimization for parallel and distributed operation and to the difficulty in vertical and horizontal scaling, impossibility of modifying a solution, e.g., due to its legal protection, etc. In order to solve these problems, an asynchronous actor model can be used that facilitates developing software for processing special computing tasks. In addition, an asynchronous actor model allows parallel processing of input tasks, provides asynchronous operation, and extends its functionality to third-party ready-made solutions, that it is impossible or forbidden to modify to meet user’s or optimization requirements. Methods, algorithms, and the entire logic of asynchronous actor operation involve transitions of actors from one state to another under the effect of input messages and returning the results of processed tasks to the frame system. This allows setting an automatic grammar and for using methods and approaches for developing software systems based on finite-state automata. The asynchronous actor operation scheme is represented by a directional state graph, which allows using the mathematical apparatus of the finite-state automaton theory (the Mealy model) to describe the same. The developed model is convenient for designing and organizing the SCT processing using computer systems. The use of the asynchronous actor model makes it possible, in special operation modes, to increase the performance of poorly optimized SCT without modifying the same, as compared to task processing without using the asynchronous actor model.

Parallel and serial task processing in the PRP paradigm: a drift\u2013diffusion model approach

Even after a long time of research on dual-tasking, the question whether the two tasks are always processed serially (response selection bottleneck models, RSB) or also in parallel (capacity-sharing models) is still going on. The first models postulate that the central processing stages of two tasks cannot overlap, producing a central processing bottleneck in Task 2. The second class of models posits that cognitive resources are shared between the central processing stages of two tasks, allowing for parallel processing. In a series of three experiments, we aimed at inducing parallel vs. serial processing by manipulating the relative frequency of short vs. long SOAs (Experiments 1 and 2) and including no-go trials in Task 2 (Experiment 3). Beyond the conventional response time (RT) analyses, we employed drift–diffusion model analyses to differentiate between parallel and serial processing. Even though our findings were rather consistent across the three experiments, they neither support unambiguously the assumptions derived from the RSB model nor those derived from capacity-sharing models. SOA frequency might lead to an adaptation to frequent time patterns. Overall, our diffusion model results and mean RTs seem to be better explained by participant’s time expectancies.

Secure and Fast Implementation of ARX-Based Block Ciphers Using ASIMD Instructions in ARMv8 Platforms

In Internet of Things services, various types of embedded devices are employed. Among them, ARM-based devices have been widely used as clients. Since these devices communicate with each other in wirelessly, transmitted data needs to be protected with secure block ciphers. Recently, several Add-Rotate-XOR (ARX)-based block ciphers, such as HIGHT and revised CHAM, have been developed for efficient encryption on embedded devices. In this paper, we present secure and fast implementations of ARX-based block ciphers HIGHT and revised CHAM in ARMv8 platforms. For performance efficiency, we basically apply task and data parallel processing mechanism by fully utilizing NEON architecture embedded in ARMv8 platforms. Typically, it is required to duplicate round key in NEON register to utilize the NEON architecture to process multiple data blocks simultaneously. In our implementations, we propose an optimal approach minimizing the cost of round key duplication and efficient key scheduling for task parallelism. For secure implementation, we develop efficient software countermeasures against realistic fault attack models. Thus, we present efficient software countermeasure based on intra-instruction redundancy. Especially, we propose enhanced random shuffling method which is the core operation for the proposed countermeasure. With the proposed random shuffling method, we can significantly reduce the overhead for preventing fault attacks. We present two versions of the software: a version providing highly fast (HF) performance without fault attack countermeasures and a version providing highly secure (HS) against fault attacks. Compared with referenced software, HF with HIGHT, revised CHAM-64/128, CHAM-128/128, and CHAM-128/256 provides about 8 times, 38 times, 13 times and 13 times of enhanced performance, respectively. Compared with previous best results having fault attack countermeasure, HS with HIGHT, revised CHAM-64/128, CHAM-128/128, and CHAM-128/256 provides about 50%, 30%, 80%, and 70% of enhanced performance, respectively. Both our HS and HF achieve better performance and higher security compared with related works.