Make Scheduling Decisions Research Articles

A discrete artificial bee colony method based on variable neighborhood structures for the distributed permutation flowshop problem with sequence-dependent setup times

• The DPFSP with SDST under minimization of total flowtime is investigated in the paper; • We propose a constructive heuristic framework, which contains three constructive heuristics, to obtain good initial solutions for our DABC vns ; • By analyzing the characteristics of problem-specific, four different neighborhood structures are proposed and selectively applied in three stages to obtain more feasible neighborhood solutions; • A new population update strategy is adopted in scout bee phase to help algorithm explore new promising area; • The experimental evaluation is used to verify the validity of constructive heuristics and DABC vns . In today's global economy, distributed manufacturing has become increasingly appropriate. Job setup time is invariably a factor in manufacturing, and needs to be considered when making scheduling decisions. This study aims to minimize the total flow time by solving a distributed permutation flowshop scheduling problem (DPFSP) with a sequence-dependent setup time (SDST) to minimize total flowtime. First, to manage the DPFSP/SDST effectively, we propose a constructive heuristic framework based on the LR heuristic, which contains three constructive heuristics to obtain good initial solutions. Second, a discrete artificial bee colony (DABC) method based on variable neighborhood structures (DABC vns ) is proposed. By analyzing the problem-specific characteristics in the DABC vns , four different neighborhood structures are proposed, which are selectively applied in three stages of DABC vns to obtain more feasible neighborhood solutions. In addition, two variable neighborhood descent algorithms are proposed, which are each based on local search methods. The algorithms are developed to improve the best solution obtained in initial population and the best solution generated in the three stages. Finally, we conduct an experiment to calibrate the parameter of the DABC vns method. To evaluate the performance of the proposed method compared to six other algorithms, comprehensive computational experiments are conducted. The simulation results indicate that the proposed method delivers the best quality solutions for the DPFSP/SDST in the minimization of total flowtime.

Designing and Implementation of Failure-Aware Based Approach for Task Scheduling in Grid Computing

Grid computing makes large-scale computations easier to handle. In heterogeneous systems like grid computing, failure is inevitable. Because of the volume and diversity of the resources, scheduling algorithm is among the most difficult challenges to overcome in grid computing. To reduce the make-span of the job to be executed a thorough understanding of scheduling in grid is important. Say there are two computing nodes that aren't being used right now. The scheduler may choose the node that has higher computing strength (for example, higher CPU speed, higher free memory), even though this node may also have high potential of failure. High potential of failure refers to the possibility of the failure occurring at execution time, resulting in the decrease of system performance. Therefore, awareness of failure is also very important in scheduling. This work proposes and implements a failure-aware scheduling method to schedule the tasks which uses both performance factors and failure factors of resources while making scheduling decision. The proposed algorithm is analyzed over various performance matrices and it shows considerably improved performance over existing algorithm

QoS-Aware Downlink Traffic Scheduling for Cellular Networks with Dual Connectivity

In a cellular network, how to preserve users’ quality of service (QoS) demands is an important issue. To provide better data services, researchers and industry have discussed the deployment of small cells in cellular networks to support dual connectivity enhancement for user equipments (UEs). By such an enhancement, a base station can dispatch downlink data to its surrounding small cells, and UEs that are located in the overlapping areas of the base station and small cells can receive downlink data from both sides simultaneously. We observe that previous works do not jointly consider QoS requirements and system capabilities when making scheduling decisions. Therefore, in this work, we design a QoS traffic scheduling scheme for dual connectivity networks. The designed scheme contains two parts. First, we propose a data dispatching decision scheme for the base station to decide how much data should be dispatched to small cells. When making a dispatching decision, the proposed scheme aims to maximize throughput and ensure that data flows can be processed in time. Second, we design a radio resource scheduling method, which aims to reduce dropping ratios of high-priority QoS data flows, while avoiding wasting radio resources. In this work, we verify our design using simulation programs. The experimental results show that compared to the existing methods, the proposed scheme can effectively increase system throughput and decrease packet drop ratios.

Representation and Reinforcement Learning for Task Scheduling in Edge Computing

Recently, many deep reinforcement learning (DRL)-based task scheduling algorithms have been widely used in edge computing (EC) to reduce energy consumption. Unlike the existing algorithms considering fixed and fewer edge nodes (servers) and tasks, in this paper, a representation model with a DRL based algorithm is proposed to adapt the dynamic change of nodes and tasks and solve the dimensional disaster in DRL caused by a massive scale. Specifically, 1) we apply the representation learning models to describe the different nodes and tasks in EC, i.e., nodes and tasks are mapped to corresponding vector sub-spaces to reduce the dimensions and store the vector space efficiently. 2) With the space after dimensionality reduction, a DRL-based algorithm is employed to learn the vector representations of nodes and tasks and make scheduling decisions. 3) The experiments are conducted with the real-world data set, and the results show that the proposed representation model with DRL-based algorithm outperforms the baselines 18.04% and 9.94% on average regarding energy consumption and service level agreement violation (SLAV), respectively.

Service scheduling for random requests with fixed waiting costs

We study service scheduling problems in a slotted system in which agents arrive with service requests according to a Bernoulli process and have to leave within two slots after arrival, service costs are quadratic in service rates, and there is also a waiting cost. We consider fixed waiting costs. We frame the problem as an average cost Markov decision process. While the studied system is a linear system with quadratic costs, it has state dependent control. Moreover, it also possesses a non-standard cost function structure rendering the optimization problem complex. Here, we characterize the optimal policy. We also consider a system in which the agents make scheduling decisions for their respective service requests keeping their own cost in view. We frame this scheduling problem as a stochastic game. Here, we provide Nash equilibrium.

A Graph Attention Mechanism-Based Multiagent Reinforcement-Learning Method for Task Scheduling in Edge Computing

Multi-access edge computing (MEC) enables end devices with limited computing power to provide effective solutions while dealing with tasks that are computationally challenging. When each end device in an MEC scenario generates multiple tasks, how to reasonably and effectively schedule these tasks is a large-scale discrete action space problem. In addition, how to exploit the objectively existing spatial structure relationships in the given scenario is also an important factor to be considered in task-scheduling algorithms. In this work, we consider indivisible, time-sensitive tasks under this scenario and formalize the task-scheduling problem to minimize the long-term losses. We propose a multiagent collaborative deep reinforcement learning (DRL)-based distributed scheduling algorithm based on graph attention neural networks (GATs) to solve task-scheduling problems in the MEC scenario. Each end device creates a graph representation agent to extract potential spatial features in the scenario and a scheduling agent to extract the timing-related features of the tasks and make scheduling decisions using a gated recurrent unit (GRU). The simulation results show that, compared with several baseline algorithms, our proposed algorithm can take advantage of the spatial positional relationship of devices in the environment, significantly reduce the average delay and drop rate, and improve link utilization.

Regret of Age-of-Information Bandits

We consider a system with a single source that measures/tracks a time-varying quantity and periodically attempts to report these measurements to a monitoring station. Each update from the source has to be scheduled on one of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula> available communication channels. The probability of success of each attempted communication is a function of the channel used. This function is unknown to the scheduler. The metric of interest is the Age-of-Information (AoI), formally defined as the time elapsed since the destination received the recent most update from the source. We model our scheduling problem as a variant of the multi-arm bandit problem with communication channels as arms. We characterize a lower bound on the AoI regret achievable by any policy and characterize the performance of UCB, Thompson Sampling, and their variants. Our analytical results show that UCB and Thompson sampling are order-optimal for AoI bandits. In addition, we propose novel policies which, unlike UCB and Thompson Sampling, use the current AoI to make scheduling decisions. Via simulations, we show the proposed AoI-aware policies outperform existing AoI-agnostic policies.

An Adaptive CMT-SCTP Scheme: A Reinforcement Learning Approach

With the continuous increase of end users and types of services, the scale of the network has shown explosive growth, which has brought tremendous pressure and challenges to network data transmission. How to achieve high-quality data transmission has become a core issue. Single-path transmission has been difficult to meet the above requirements. The concurrent multipath transfer extension for stream control transmission protocol (CMT-SCTP), which supports multipath and independent data streams, can solve this problem. However, the current transmission path assessment scheme has too large granularity to make full use of the resources of the transition zone. Most studies ignore the different requirements of different services, a single transmission strategy, and the lack of an intelligent dynamic adjustment mechanism. Therefore, we designed a QCMT(Q-learning based CMT-SCTP) scheduling method. This method considers the multi-dimensional characteristics of the path and the characteristic preferences of different services, periodically evaluates and trains the reinforcement learning model for service adaptation, and makes scheduling decisions dynamically. Experimental results show that dynamic scheduling based on path parameters and service preferences can reduce message delay and improve network throughput.

A Fuzzy Logic-Based On-Demand Charging Algorithm for Wireless Rechargeable Sensor Networks With Multiple Chargers

Mobile chargers have greatly promoted the wireless rechargeable sensor networks (WRSNs). While most recent works have focused on recharging the WRSNs in an on-demand fashion, little attention has been paid on joint consideration of multiple mobile chargers (MCs) and multi-node energy transfer for determining the charging schedule of energy-hungry nodes. Moreover, most of the schemes leave out the contemplation of multiple network attributes while making scheduling decisions and even they overlook the issue of ill-timed charging response to the nodes with uneven energy consumption rates. In this paper, we address the aforesaid issues together and propose a novel scheduling scheme for on-demand charging in WRSNs. We first present an efficient network partitioning method for distributing the MCs so as to evenly balance their workload. We next adopt the fuzzy logic which blends various network attributes for determining the charging schedule of the MCs. We also formulate an expression to determine the charging threshold for the nodes that vary depending on their energy consumption rate. Extensive simulations are conducted to demonstrate the effectiveness and competitiveness of our scheme. The comparison results reveal that the proposed scheme improves charging performance compared to the state-of-the-art schemes with respect to various performance metrics.

Adaptive Task Partitioning for Performance Evaluation in Cluster based Heterogeneous Environments

Due to continual server replacement, datacenter-scale clusters are developing toward heterogeneous hardware designs. Meanwhile, datacenters are frequently used by a variety of users for a variety of purposes. Due to multi-tenant interferences, it frequently exhibits high performance heterogeneity. When contrasted to in-house dedicated clusters, deploying MapReduce on such heterogeneous clusters poses major hurdles in attaining adequate application performance. Heterogeneity can cause significant performance degradation in job execution, despite current optimizations on task scheduling and load balancing, because most MapReduce implementations were developed for homogeneous contexts. To make scheduling decisions, the majority of extant adaptive strategies assume a priori knowledge of particular job characteristics. However, without spending a significant cost, such information is not readily available. The suggested framework Adaptive Control Self-tuning provides a significant improvement over existing methods at moderate to high system utilizations, according to the evaluation results.

Budget‐aware scheduling algorithms for scientific workflows with stochastic task weights on infrastructure as a service Cloud platforms

SummaryThis paper introduces several budget‐aware algorithms to deploy scientific workflows on Infrastructure as a Service Cloud platforms, where users can request Virtual Machines (VMs) of different types, each with specific cost and computing resources. We use a realistic application/platform model with stochastic task weights, and VMs communicating through a Cloud storage. We extend two well‐known algorithms, MinMin and HEFT, and make scheduling decisions based upon machine availability and remaining budget. During the mapping process, the budget‐aware algorithms make conservative assumptions to avoid exceeding the initial budget; we further improve the results with refined versions that aim at rescheduling some tasks onto faster VMs, thereby spending any budget fraction leftover by the first allocation. These refined variants are much more time‐consuming than the former algorithms, so there is a trade‐off to find in terms of scalability. We report an extensive set of simulations with workflows from the Pegasus benchmark suite. Most of the time, our budget‐aware algorithms succeed in achieving efficient makespans while enforcing the given budget, and this despite the uncertainty in task weights.

Age-Optimal Scheduling for Heterogeneous Traffic With Timely Throughput Constraints

We consider a base station supporting two types of traffics, i.e., status update traffic and timely throughput traffic. The goal is to improve the information freshness of status update traffic while satisfying timely throughput constraints. Age of Information (AoI) is adopted as a metric for information freshness. We first propose an age-aware policy that makes scheduling decisions based on the current value of AoI directly. Given timely throughput constraint, an upper bound of the weighted average AoI under this policy is provided. To evaluate policy performance, it is important to obtain the minimum weighted average AoI achievable given timely throughput constraint. A low complexity method is proposed to estimate a lower bound of this value. Furthermore, inspired by the estimation procedure, we design an age-oblivious policy that does not rely on the current AoI to make scheduling decisions. Surprisingly, simulation results show that the weighted average AoI of the age-oblivious policy is comparable to that of the age-aware policy, and both are close to the lower bound.

Energy Efficient Scheduling Algorithm for Structural Health Building Monitoring System(Shbm) to Increase the Battery Lifetime

A structural health building monitoring system is invented, achieved and attempted using Wireless Sensor Networks .Ambient vibration of the structure is audited and is implemented to identify the health status of the structure. With WSN, low cost monitoring is possible without intruding with the action of the structure. In this cluster head is heavily loaded since it gathers information from all other member nodes and integrate then and forward to sink either directly or through intermediate other cluster head. I near future cluster head drain and we need to select another cluster head via re clustering. The re clustering also led to energy loss to select the cluster head based on weight age value.WSN consumes less power and it is used to track slow targets .WSNs covers only small distance. WSN has high failure rate and we have to plant excessive nodes. Hence redundant nodes dissipate more power. To avoid this, sleep mode and scheduling algorithm is used in this paper. The objective of the paper is to design a scheduling algorithm to trace data with target rate. The proposed scheme is a kind of an adaptive/Cyclic on–off scheduling scheme in where sensors nodes use only local information to make scheduling decisions.

AN INTEGRATED FRAMEWORK FOR SLA-AWARE MULTI-OBJECTIVE TASK SCHEDULING IN CLOUD COMPUTING

Cloud services are offered to consumers based on Service Level Agreements (SLAs) signed between Cloud Service Provider (CSP) and consumer. Due to on-demand provisioning of resources there is exponential growth of cloud consumers. Job scheduling is one of the areas that has attracted researchers to improve performance of cloud management system. Along with the on premise infrastructure, Small and Medium Enterprises (SMEs) also depend on public cloud infrastructure (leading to hybrid cloud) for seamless continuity of their businesses. In this context, ensuring SLAs and effective management of hybrid cloud resources are major challenging issues to be considered. Hence, there is a need for an effective scheduling algorithm which considers multiple objective functions like SLA (deadline), cost and energy while making scheduling decisions. Most of the state of the art schedulers in hybrid cloud environment considered single objective function. However, in real world, it is inadequate for scheduling effectiveness. To overcome this problem, we proposed an integrated framework which ensures SLAs (deadline), cost effectiveness and energy efficiency with an underlying scheduling algorithm known as SCE-TS. This algorithm is evaluated with different workloads and SLAs using a cloud platform. The empirical study revealed that the proposed framework improves scheduling efficiency in terms of meeting SLAs, cost and energy efficiency. It is evaluated and compared with the state of the art and found to be effective in making scheduling decisions in cloud environment.

Deep Reinforcement Learning for Scheduling in an Edge Computing‐Based Industrial Internet of Things

The demand for improving productivity in manufacturing systems makes the industrial Internet of things (IIoT) an important research area spawned by the Internet of things (IoT). In IIoT systems, there is an increasing demand for different types of industrial equipment to exchange stream data with different delays. Communications between massive heterogeneous industrial devices and clouds will cause high latency and require high network bandwidth. The introduction of edge computing in the IIoT can address unacceptable processing latency and reduce the heavy link burden. However, the limited resources in edge computing servers are one of the difficulties in formulating communication scheduling and resource allocation strategies. In this article, we use deep reinforcement learning (DRL) to solve the scheduling problem in edge computing to improve the quality of services provided to users in IIoT applications. First, we propose a hierarchical scheduling model considering the central‐edge computing heterogeneous architecture. Then, according to the model characteristics, a deep intelligent scheduling algorithm (DISA) based on a double deep Q network (DDQN) framework is proposed to make scheduling decisions for communication. We compare DISA with other baseline solutions using various performance metrics. Simulation results show that the proposed algorithm is more effective than other baseline algorithms.

An evolutionary fuzzy scheduler for multi-objective resource allocation in fog computing

With rapid development of the Internet of Things (IoT), a vast amount of raw data produced by IoT devices needs to be processed promptly. Compared to cloud computing, fog computing nodes are closer to data resource for decreasing the end-to-end transmission latency. Considering the limited resource of IoT devices, offloading computationally-intensive tasks to the servers with high computing capability is essential in the IoT–fog–cloud system to complete those tasks on time. In this work, we propose a fuzzy logical offloading strategy for IoT applications characterized by uncertain parameters to optimize both agreement index and robustness. A multi-objective Estimation of Distribution Algorithm (EDA) is designed to learn and optimize the fuzzy offloading strategy from a diversity of the applications. The algorithm partitions applications into independent clusters, so that each cluster can be allocated to the corresponding tier for further processing. Thus, system resources are saved by making scheduling decisions in a reduced search space. Simulation studies on benchmark problems and real-world cases are carried out to verify the efficiency of our proposed algorithm. Pareto sets produced by our algorithm outperformed classic heuristic solutions for 88.3% benchmark cases and dominated Pareto sets of two state-of-art multi-objective algorithms for 92.7% and 94.4% cases correspondingly.

An Information Theory Inspired Real-Time Self-Adaptive Scheduling for Production-Logistics Resources: Framework, Principle, and Implementation.

The development of industrial-enabling technology, such as the industrial Internet of Things and physical network system, makes it possible to use real-time information in production-logistics scheduling. Real-time information in an intelligent factory is random, such as the arrival of customers’ jobs, and fuzzy, such as the processing time of Production-Logistics Resources. Besides, the coordination of production and logistic resources in a flexible workshop is also a hot issue. The availability of this information will enhance the quality of making scheduling decisions. However, when and how to use this information to realize the adaptive collaboration of Production-Logistics Resources are vital issues. Therefore, this paper studies the above problems by establishing a real-time reaction scheduling framework of Production-Logistics Resources dynamic cooperation. Firstly, a real-time task triggering strategy to maximize information utilization is proposed to explore when to use real-time information. Secondly, a collaborative method for Production-Logistics Resources is studied to explore how to use real-time information. Thirdly, a real-time self-adaptive scheduling algorithm based on information entropy is utilized to obtain a stable and feasible solution. Finally, the effectiveness and advancement of the proposed method are verified by a practical case.

Learning scheduling bursty requests in Mobile Edge Computing using DeepLoad

The emergence of Mobile Edge Computing (MEC) alleviates the large transmission latency resulting from the traditional cloud computing. For the compute-intensive requests such as video analysis, mobile users prefer to obtain a desired quality of experience (QoE) with neglected latency and reduced energy consumption. The popularity of smart devices allows users to release a run of compute-intensive as well as latency-sensitive requests anywhere, which may lead to bursty requests. A single resource-constrained edge server nearby is capable of handling a small amount of requests quickly, yet it seems helpless when encountering bursty compute-intensive requests. Despite the abundance of recently proposed schemes, the majority focus on efficiently scheduling pending requests in a single edge server, and ignored the potential role of edge collaboration to schedule bursty requests. Besides, while some recent studies proposed to finish a task using multiple devices, they focused on collaboration between mobile devices rather than between edge servers. Hence, we propose DeepLoad, a S2S system that schedules the bursty requests with a collaborative method using reinforcement learning (RL). DeepLoad decouples the scheduling decision into AP selection for setting the access point and workload redistribution for collaborative servers. DeepLoad trains a neural network model that picks decisions for each request based on observations collected by mobile devices. DeepLoad learns to make scheduling decisions solely through the resulting performance of historical decisions rather than rely on pre-programmed models or specific assumptions for the environment. Naturally, DeepLoad automatically learns the scheduling algorithm for each request and obtains a gratifying QoE. We aim to maximize the fraction of requests finished before their attached deadlines. Based on the Shanghai taxi trajectory data set, we design a simulator to obtain abundant samples, and leverage two GeForce GTX TITAN Xp GPUs to train the Actor–Critic network. Compared to the state-of-the-art bandwidth-based and server resources-based methods, DeepLoad can achieve a significant improvement in average fraction.

Less is More: Service Profit Maximization in Geo-Distributed Clouds

Nowadays cloud providers purchase a good deal of bandwidth from Internet service providers to satisfy the growing requests from corporate customers for the exclusive use of inter-datacenter bandwidth. For exclusive bandwidth services, neither maximizing the revenue nor minimizing the cost can bring the maximal profit to cloud providers. The diversity of bandwidth prices and the random arrival time of user requests further increase the difficulty in economically scheduling the services to meet user requests from cloud providers. In this article, we propose to help cloud providers maximize their service profits by properly selecting user requests to serve rather than satisfying them all. We formulate the problem of service profit maximization and prove its NP-hardness. To handle offline request submission, we propose a solution that maximizes the service profit by alternately maximizing the service revenue and minimizing the service cost. To maximize service profit under online request submission, we propose an online scheduling algorithm that carefully handles the risk of not being able to pay off the incremental service cost and makes scheduling decisions in real time. Our extensive evaluations demonstrate that our solutions can achieve more than 1.6x the service profits of existing solutions.

Managing uncertainty in demand through inventory and workforce development

This paper studies whether and how a manufacturer can mitigate the effects of demand uncertainty during production ramp up by jointly managing inventory and capacity development, determined by the productivity of a heterogeneous workforce. One way the manufacturer can direct its capacity development is by recognising that workers become more productive as they gain experience when making scheduling decisions. We consider a situation in which demands for multiple products are periodically revealed and a manufacturer must decide what each individual does in the production process, how much inventory is accumulated, and how demands are met. To study the potential benefits, we formulate a Markov decision process model which we solve using a one-step lookahead approximation. Results of an extensive computational study show that, given the effectiveness associated with both directing capacity and holding inventory, manufacturers should consider the workforce's capacity to learn when making job assignments during production ramp up. Analysis of the results suggests that inventory and capacity building patterns depend on product value and the workforce's experience levels. Further analysis shows that slower learners and those with the potential to achieve higher productivity should specialise more. Faster learners, on the other hand, are best suited for cross training.