Parallel Scheduling Research Articles

Retraction Note: Multi-level parallel scheduling of dependent-tasks using graph-partitioning and hybrid approaches over edge-cloud

Retraction Note: Multi-level parallel scheduling of dependent-tasks using graph-partitioning and hybrid approaches over edge-cloud

Research on Parallel Task Scheduling Algorithm of SaaS Platform Based on Dynamic Adaptive Particle Swarm Optimization in Cloud Service Environment

To efficiently realize the parallel task scheduling of SaaS platform in large-scale cloud service environment, this paper studies the parallel task scheduling algorithm of SaaS platform based on dynamic adaptive particle swarm optimization in cloud service environment. Users access the cloud through the user access interface module, and issue task scheduling instructions or send task scheduling requests. After the service management module provides diversified application service support according to the scheduling requirements, the core service module determines the SaaS platform parallel scheduling objective function, and uses dynamic adaptive particle swarm optimization to solve the objective function to obtain the SaaS platform parallel task scheduling results. The test results show that the algorithm has better multi-objective solving ability and can obtain higher quality objective solutions, and the test results of the total execution time of parallel scheduling tasks and the total transmission time of task data on SaaS platform are all within 30 s. The results of virtual machine resource load balancing degree are all below 15%; the utilization rate of virtual machine resources is above 92.2%.

A Parallel Optimization Method for Robustness Verification of Deep Neural Networks

Deep neural networks (DNNs) have gained considerable attention for their expressive capabilities, but unfortunately they have serious robustness risks. Formal verification is an important technique to ensure network reliability. However, current verification techniques are unsatisfactory in time performance, which hinders the practical applications. To address this issue, we propose an efficient optimization method based on parallel acceleration with more computing resources. The method involves the speedup configuration of a partition-based verification aligned with the structures and robustness formal specifications of DNNs. A parallel verification framework is designed specifically for neural network verification systems, which integrates various auxiliary modules and accommodates diverse verification modes. The efficient parallel scheduling of verification queries within the framework enhances resource utilization and enables the system to process a substantial volume of verification tasks. We conduct extensive experiments on multiple commonly used verification benchmarks to demonstrate the rationality and effectiveness of the proposed method. The results show that higher efficiency is achieved after parallel optimization integration.

Adaptive Parallel Scheduling Scheme for Smart Contract

With the increasing demand for decentralized systems and the widespread usage of blockchain, low throughput and high latency have become the biggest stumbling blocks in the development of blockchain systems. This problem seriously hinders the expansion of blockchain and its application in production. Most existing smart contract scheduling solutions use static feature analysis to prevent contract conflicts during parallel execution. However, the conflicts between transactions are complex; static feature analysis is not accurate enough. In this paper, we first build the dependency between smart contracts by analyzing the features. After numerous experiments, we propose a conflict model to adjust the relationship between threads and conflict to achieve high throughput and low latency. Based on these works, we propose adaptive parallel scheduling for smart contracts on the blockchain. Our adaptive parallel scheduling can distinguish conflicts between smart contracts and dynamically adjust the execution strategy of smart contracts based on the conflict factors we define. We implement our scheme on ChainMaker, one of the most popular open-source permissioned blockchains, and build experiments to verify our solution. Regarding latency, our solution demonstrates remarkable efficiency compared with the fully parallel scheme, particularly in high-conflict transaction scenarios, where our solution achieves latency levels just one-twentieth of the fully parallel scheme. Regarding throughput, our solution significantly outperforms the fully parallel scheme, achieving 30 times higher throughput in high-conflict transaction scenarios. These results highlight the superior performance and effectiveness of our solution in addressing latency and throughput challenges, particularly in environments with high transaction conflicts.

Real-Time Embedded Simulation Platform for Hippocampal Traveling Waves of Electric Field Conduction

The investigation of hippocampal traveling waves has gained significant importance in comprehending and treating neural disorders such as epilepsy, as well as unraveling the neural mechanisms underlying memory and cognition. Recently, it has been discovered through both in vivo and in vitro experiments that hippocampal traveling waves are typically characterized by the coexistence of fast and slow waves. However, electrophysiological experiments face limitations in terms of cost, reproducibility, and ethical considerations, which hinder the exploration of the mechanisms behind these traveling waves. Model-based real-time virtual simulations can serve as a reliable alternative to pre-experiments on hippocampal preparations. In this paper, we propose a real-time simulation method for traveling waves of electric field conduction on a 2D plane by implementing a hippocampal network model on a multi-core parallel embedded computing platform (MPEP). A numerical model, reproducing both NMDA-dependent fast waves and Ca-dependent slow waves, is optimized for deployment on this platform. A multi-core parallel scheduling policy is employed to address the conflict between model complexity and limited physical resources. With the support of a graphical user interface (GUI), users can rapidly construct large-scale models and monitor the progress of real simulations. Experimental results using MPEP with four computing boards and one routing board demonstrate that a hippocampal network with a 200 × 16 pyramidal neuron array can execute real-time generation of both fast and slow traveling waves with total power consumption below 500 mW. This study presents a real-time virtual simulation strategy as an efficient alternative to electrophysiological experiments for future research on hippocampal traveling waves.

A dynamic simulation framework based on hybrid modeling paradigm for parallel scheduling systems in warehouses

Efficient warehouse management is of crucial significance for the smooth operation of a company's supply chain. With the challenging nature of warehouse environment changes, research on warehouse operational issues is increasingly important. Moreover, system simulation has emerged as a prevalent means of investigating warehouse operational management. However, prior research on warehouse simulation either utilized singular paradigms or lacked a generalized modeling framework. This remains a challenge for modelers exploring diverse domains of warehouse-related issues using simulation techniques. In this context, this study presents an integrated control methodology(ICM) based on the state changes of dispatchers and logistics equipments, the discrimination of task scenarios, and the behaviors of dispatchers. This methodology is incorporated into the warehouse workflow model. The workflow model serves as a conceptual abstraction of a warehouse's parallel scheduling system, while the integrated control methodology (ICM) simulates the entire decision-making process of dispatchers to resolve potential deadlock issues during task execution. Subsequently, we utilize a steel slab warehouse in a case study, employing a dynamic simulation using a hybrid paradigm based on Discrete Event Simulation (DES) and Agent-Based Simulation (ABS) to replicate the historical scheduling process within the warehouse. This demonstration confirms the feasibility of the proposed framework. Finally, we devise multiple dimensions of validation metrics to confirm the model's effectiveness.

Eventually-consistent federated scheduling for data center workloads

Data center schedulers operate at unprecedented scales today to accommodate the growing demand for computing and storage power. The challenge that schedulers face is meeting the requirements of scheduling speeds despite the scale. To do so, most scheduler architectures use parallelism. However, these architectures consist of multiple parallel scheduling entities that can only utilize partial knowledge of the data center’s state, as maintaining consistent global knowledge or state would involve considerable communication overhead. The disadvantage of scheduling without global knowledge is sub-optimal placements—tasks may be made to wait in queues even though there are resources available in zones outside the scope of the scheduling entity’s state. This leads to unnecessary queuing overheads and lower resource utilization of the data center.In this paper, we extend our previous work on Thiyyakat et al. (2023), a federated decentralized data center scheduling architecture that uses eventual consistency. The architecture utilizes both parallelism and an eventually consistent global state in each of its scheduling entities to make fast decisions in a scalable manner. In our work, we compare Megha with 3 scheduling architectures — Sparrow, Eagle, and Pigeon, using simulation. We also evaluate Megha’s prototype on a 123-node cluster and compare its performance with Pigeon’s prototype using cluster traces. The results of our experiments show that Megha consistently reduces delays in job completion time when compared to other architectures.

Weighted Multi-Skill Resource Constrained Project Scheduling: A Greedy and Parallel Scheduling Approach

Weighted Multi-Skill Resource Constrained Project Scheduling: A Greedy and Parallel Scheduling Approach

Efficient parallel scheduling with power control and successive interference cancellation in wireless sensor networks

In wireless sensor networks (WSNs), spectrum efficiency and energy efficiency are two challenging issues to be addressed due to the scarcity of spectrum and its limited energy. In this paper, we study efficient parallel scheduling with power control and successive interference cancellation (SIC) aimed at maximizing spectrum and energy efficiency, and formulate the efficiency maximization scheduling (EMS) problem as a mixed integer nonlinear programming (MINIP) problem. Given the high complexity caused by mixed variables, we first analyze the critical variables affecting efficiency to transform the MINIP problem into a combinatorial linear programming (LP) problem, which is further decomposed into two linear minimization problems. Then, a recursive relationship between the optimization variables is explored, through which we propose a two-stage scheme that solves the minimization problems with lower complexity. Finally, distributed implementation of the scheduling scheme is provided to iteratively explore the maximum feasible concurrent links in the networks. Extensive results demonstrate the superior performance of our proposed scheme in improving spectrum and energy efficiency. Specifically, our scheme achieves 12.16% to 66.15% higher spectrum efficiency and 18.84% to 88.20% higher energy efficiency than the existing schemes.

APapo: An asynchronous parallel optimization method for DNN models

To address the challenges related to segmentation complexity, high memory usage, extended training duration, and low equipment utilization in parallel optimization of large-scale deep neural network (DNN) models, this paper proposes an asynchronous parallel optimization method APapo. Firstly, a multi-iteration asynchronous pipeline parallel scheduling was established for model parallel computing tasks, controlling the specific scheduling process of micro-batch units to address gradient delay updating during asynchronous iteration. Secondly, combined with the given network model and hardware configuration, a dynamic programming strategy for computing resources and model tasks was designed to achieve dynamic segmentation of model computing tasks and optimal matching of computing resources. Finally, an optimization strategy for runtime scheduling of computing resources and model tasks was developed, using improved device streams to maximize the overlap between computing and communication, thus improving the utilization rate of computing resources and reducing training time. Experimental results show that the APapo method achieves fine-grained task segmentation, maximizes the utilization rate of each GPU computing resource, and on average improves the training speed of large-scale deep neural network models by 2.8 times while maintaining the training accuracy of the model compared to existing parallel optimization methods.

High Throughput VMs Placement With Constrained Communication Overhead and Provable Guarantees

Placement of VMs in the cloud is one of the most fundamental problems in systems research. Traditionally, placement algorithms assume that the schedulers have complete information about the currently available resources at each host. However, this assumption is in many cases unrealistic, as gathering fresh status information from each of the thousands of hosts in a large data center incurs excessive communication overhead, which results in long queueing delays. Efforts to resolve this problem by employing several parallel schedulers typically exhibit collisions when several schedulers are simultaneously trying to place VMs on the same host. Our work analyzes the performance of various placement algorithms and provides empirical evidence that using multiple randomized schedulers obtains high throughput, while significantly decreasing both the communication overhead, and the number of collisions between schedulers. We, therefore, introduce Adaptive Partial State Random (APSR) – an efficient parallel random resource management algorithm that samples only from a small number of hosts and dynamically adjusts the degree of parallelism to provide provable guarantees on the probability of collisions between distinct schedulers. We formally analyze APSR, evaluate it on real workloads, and integrate it into the popular OpenStack cloud management platform. Our evaluation shows that APSR matches the throughput provided by other parallel schedulers, while achieving up to 13x lower decline ratio and a reduction of over 85% in communication overheads.

Efficient Techniques for Residential Appliances Scheduling in Smart Homes for Energy Management Using Multiple Knapsack Problem

The evolution of the smart grid has enabled residential users to manage the ever-growing energy demand in an efficient manner. The smart grid plays an important role in managing this huge energy demand of residential households. A home energy management system enhances the efficiency of the energy infrastructure of smart homes and provides an opportunity for residential users to optimize their energy consumption. Smart homes contribute significantly to reducing electricity consumption costs by scheduling domestic appliances effectively. This residential appliance scheduling problem is the motivation to find an optimal appliance schedule for users that could balance the load profile of the home and helps in minimizing electricity cost (EC) and peak-to-average ratio (PAR). In this paper, we have focused on appliance scheduling on the consumer side. Two novel home energy management models are proposed using multiple scheduling options. The residential appliance scheduling problem is formulated using the multiple knapsack technique. Serial and parallel scheduling algorithms of home appliances namely MKSI (Multiple knapsacks with serial implementation) and MKPI (Multiple knapsacks with parallel implementation) are proposed to reduce electricity cost and PAR. Price-based demand response techniques are incorporated to shift appliances from peak hours to off-peak hours to optimize energy consumption. The proposed algorithms are tested on real-time datasets and evaluated based on time of use pricing tariff and critical peak pricing. The performance of both the algorithms is compared with the unscheduled scenario and existing algorithm. Simulations show that both proposed algorithms are efficient methods for home energy management to minimize PAR and electricity bills of consumers. The proposed MKSI algorithm achieves cost reduction of 20.26% and 42.53% for TOU and CPP, respectively as compared to the unscheduled scenario while PAR is reduced by 45.07% and 39.51% for TOU and CPP, respectively. The proposed MKPI algorithm achieves 22.33% and 46.36% cost reduction compared to the unscheduled case for TOU and CPP while the PAR ratio is reduced by 46.47% and 41.16% for TOU and CPP respectively.

A priority rule for scheduling shared due dates in the resource-constrained project scheduling problem

In resource-constrained project scheduling, project milestones, rental equipment, contracted labor, and product development checkpoints all conceptually impose shared due dates, which expand the concept of conventional due dates by permitting their assignment to sets of activities and by allowing multiple shared due date assignments, or none at all, to a single activity. Existing heuristic algorithms in the literature that solve the resource-constrained project scheduling problem are agnostic to shared due dates and cannot utilize information about their assignments in making scheduling decisions. To this end, we propose a novel priority rule compatible with both serial and parallel schedule generation schemes that integrates shared due date information into the heuristic scheduling decisions. We evaluate the effectiveness of our priority rule by demonstrating its dominance over existing priority rules in the literature and by comparing its performance with an exact algorithm. In comparing with the latter, we demonstrate both the computational efficiency of the heuristic as well as the quality of the solutions it produces.

Automatic Pipeline Parallelism: A Parallel Inference Framework for Deep Learning Applications in 6G Mobile Communication Systems

With the rapid development of wireless communication, achieving the neXt generation Ultra-Reliable and Low-Latency Communications (xURLLC) in 6G mobile communication systems has become a critical problem. Among many applications in xURLLC, deep learning model inference requires improvement over its efficiency. Due to the heterogeneous hardware environment in 6G, parallel schedules from distributed machine learning and edge computing has been borrowed to tackle the efficiency problem. However, traditional parallel schedules suffer from high latency, low throughput, and low device utility. In this paper, we propose Automatic Pipeline Parallelism ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">AP</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ), a parallel inference framework for deep learning applications in 6G mobile communication systems, to improve the model inference efficiency while maintaining reliability. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">AP</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> contains three sub-modules. A task-device affinity predictor predicts a task’s expected execution time on a given device. The parallel inference arrangement optimizer finds the most suitable device for each task. The parallel inference scheduler converts the arrangement to a schedule that can be directly executed in the system. The experimental results show that <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">AP</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> can achieve better latency, throughput, reliability, and device utility than other parallel schedules. Also, the priority of the sub-module designs has been approved through the experiments.

Optimizing Berth Allocation in Maritime Transportation with Quay Crane Setup Times Using Reinforcement Learning

Maritime transportation plays a critical role in global trade as it accounts for over 80% of all merchandise movement. Given the growing volume of maritime freight, it is vital to have an efficient system for handling ships and cargos at ports. The current first-come-first-serve method is insufficient in maintaining operational efficiency, especially under complicated conditions such as parallel scheduling with different cargo setups. In addition, in the face of rising demand, data-driven strategies are necessary. To tackle this issue, this paper proposes a mixed-integer model for allocating quay cranes, terminals, and berths. It considers not only cargo types, but also the time required for a quay crane setup. The proposed model features a greedy-insert-based offline algorithm that optimizes berth allocation when vessel information is available. In situations where vessel information is uncertain, the model utilizes an online optimization strategy based on a reinforcement-learning algorithm that is capable of learning from feedback and of adapting quickly in real time. The results of the numerical experiments demonstrate that both the offline and online algorithms can significantly enhance cargo handling efficiency and overall harbor operation. Furthermore, they have the potential to be extended to other complex settings.

A heuristic solution framework for the resource-constrained multi-aircraft scheduling problem with transfer of resources and aircraft

Efficient and feasible flight deck operation scheduling plan is of great significance for improving the sortie rate of carrier aircraft. Oxygen filling, fueling, and other operations of each carrier aircraft are executed in the pre-flight preparation stage. Different personnel, support equipments, and supply resources, are required while different operations are being executed. To date, most approaches proposed in the literature are based on the unrealistic assumption that the transfer process of carrier aircraft can be ignored. To close this gap, the resource-constrained multi-aircraft scheduling problem on the flight deck is examined in this study. The main contributions of the study are as follows. First, by considering the transfer process of aircraft on the flight deck operations, the model formulation of resource-constrained multi-aircraft scheduling problem is established. Second, the parallel and serial scheduling framework are proposed, where parking spot priority rules and resource allocation rules are presented to select the most suitable resources. Third, a modified genetic programming algorithm is presented for obtaining efficient activity priority rules. Fourth, the well-performed scheduling framework, priority rules, and allocation rules are selected by conducting computational experiments. Furthermore, experiment results show that the parallel scheduling framework outperforms the serial framework in most experiments (average outperformance is 10.19%), and the parking spot priority rules slightly affect the serial scheduling framework.

Unrelated parallel machine scheduling models with machine availability and eligibility constraints

In this paper, we address the unrelated parallel machine scheduling problem with calendar, which consists of a generic intermittent availability scheme with non-regular active and inactive time windows in sequence. The problem additionally considers machine eligibility and release date constraints, as well as the possibility of pausing or not the processing of the jobs between successive active time windows. This problem is quite frequent and relevant in several practical settings in which it is necessary to schedule parallel resources with different calendars, such as different weekly and daily shifts, non-regular not worked days, holidays, and preventive maintenance periods, among others, while considering the resources eligibility. We present novel mixed integer linear programming models for this problem with the aim of minimizing the makespan. Basic formulations of these models are presented as well as tight lower bounds for the objective, including enhanced formulations. To the best of our knowledge, there are no other works that tackled this problem and that proposed optimization models with the level of detail and generality here considered. The models presented are evaluated using a state-of-the-art general-purpose solver, and the computational tests are performed on a variety of instances. The results validate the models and show that they are can handle instances in realistic situations where the problem occurs. The results also show that the enhanced formulations are able to find several optimal solutions, and that within acceptable computational times they are able to sharply reduce the optimality gaps when compared to the more basic formulations.

Preface

The 2nd Agrifood System International Conference (ASIC) Professor Jurnalis Kamil Convention Hall, Padang, West Sumatra, Indonesia, 8-9 November 2022“Research advancement and innovations in agroecology and smart agrifood systems.”The 2nd Agrifood System International Conference (ASIC 2022) was successfully held on 8-9 November 2022. Due to the covid-19 pandemic, this event was held virtually via the zoom platform, directly from Professor Jurnalis Kamil Convention Hall, Padang, West Sumatra, Indonesia. This event was organized by the Faculty of Agriculture, Universitas Andalas, Indonesia, and became a part of the event to commemorate the 68th anniversary of the faculty. The theme of the ASIC 2022 was: “Research advancement and innovations in agroecology and smart agrifood systems.”There have been numerous revolutions in agriculture, which have improved competency and led to record-breaking yields and gains. The latest process is “smart farming,” contributing to humanity’s survival and future prosperity. Smart farming presents numerous prospects for pervasive interconnection and database computer technology as part of Industry 4.0. Smart farming is the idea of agricultural practice in a creative manner while utilizing cutting-edge technology to improve the quantity and quality of agricultural goods. New methods to assure global food safety are part of the future of the food manufacturing industry. It enables farmers to boost yields more effectively and efficiently. Fertilizers, labor, seeds, and water are just a few resources that can be saved. Smart farming has supporting applications, including land management, selection of varieties, minimizing synthetic fertilizers and pesticide inputs, and replacing them with environmentally friendly inputs. Research and related technological innovations have been carried out but have yet to be adopted and properly integrated.The main objective of this conference was to provide a venue for exchanging knowledge, scientific advancement, and innovative ideas among researchers, academicians, governments, and organizations. The scope includes plant breeding and crop production, soil management, plant protection and food safety, the socio-economic of agriculture and natural resources, and all topics related to agriculture. The committee received more than two hundred paper abstracts coming from 46 institutions, national and international. We encourage student presenters from undergraduate to doctoral programs to present their papers; hence, around 25% of abstracts come from them.The conference program was divided into two main segments: plenary and parallel. The plenary session invited 13 speakers from within and outside the country and was attended by 610 participants during the two days’ activities. On behalf of the committee, we greatly appreciate the seven speakers contributing and sharing their knowledge at this event: Dr. Silvain R Perret, Scientific Director of CIRAD, France; Mr. Pierre Ferrand from FAO, Regional Office for Asia and the Pacific; Prof. Norman Uphoff, SRI Scientist from Cornell University, USA; Dr. Jauhar Ali, Rice hybrid breeder from IRRI, Philippines; Dr. Trevor A. Jackson, Plant protection scientist from IAPPS/ Coordinator Region XII; Prof. Shamshuddin Jusop, Soil Science Scientist from UPM, Malaysia; and Dr. Wahono: Drone creator from UMM, Indonesia. We also introduced five invited speakers from the Faculty of Agriculture: Dr. Irawati Chaniago - Crop Production; Dr. Dini Hervani - Plant Breeding; Dr. Eka Candra Lina - Plant Protection; Dr. Yuerlita - Socio-economics of Agriculture; Dr. Hery Bachrizal Tanjung - Agricultural Extension. In addition, we have provided an online workshop conducted as a side event on successfully publishing an article in IOP-EES Proceeding.Finally, let me express my sincere gratitude to all presenters, participants, and committee members who contributed significantly to this event’s success. Special thanks go to the Rector of Universitas Andalas and the head of the research institute and community service of Universitas Andalas for all the support during the event. We hope to deliver the 3rd ASIC in 2024.Warmest regards, Dr. My Syahrawati Chairperson of the Organizing Committee List of Documentations, Conference Committee, Conference Schedule, Parallel Schedule, List of Presenters are available in this Pdf.

Game Theory Approach for Concurrent Delay Analysis

Concurrent delays are multiple project participants incurring delays on parallel schedule paths. They have been controversial because it is challenging to apportion liability for liquidated damages fairly and equitably. To date, there exist neither a transparent solution, nor one that addresses all possible scenarios, including mixed, where some participants are ahead of schedule and others are behind. This study therefore proposes a comprehensive model that also newly includes incentive bonuses. This cooperative game theory–based approach has three steps: (1) identify as-planned and as-built schedule inputs; (2) categorize delayed and ahead activities; and (3) operate games to determine Shapley values. Compared with prior delay analysis methods, our model is more robust and apportions liability not just in days, but also in financial terms. A real-world case was analyzed to validate it and demonstrate the reasonableness of its results. This study contributes to the body of knowledge by creating an objective concurrent delay analysis using readily available project data. It also aids in incentivizing project participants to deliver their best possible performance.

Parallel Scheduling of Large-Scale Tasks for Industrial Cloud–Edge Collaboration

Industrial Internet of Things is moving toward an intelligent level with large-scale collaborative cloud and edge resources, making it possible for online supervision, fast analysis, and precise control for many manufacturing job shops. However, online processing of large-scale industrial computation brings huge communication overhead and energy consumption among cloud, edge, and end devices. To improve the performance of the cloud–edge collaboration, this article establishes a practical model of task scheduling considering two kinds of cloud–edge collaborative modes. We propose a parallel group-merge evolutionary algorithm to assign thousands of tasks in seconds. The algorithm separates tasks into weakly correlated groups and applies modified evolutionary operators to find a subsolution for each group. Then, the subsolutions are merged to form a complete solution for fine-tuning based on the cross-use of heuristics. Experimental results show that the proposed method could assign thousands of tasks to cloud servers and edge servers in seconds, reduce the overall task computing time by 36.97%, and save the overall energy by 23.71% at most.