Task scheduling, resource provisioning, and load balancing on scientific workflows using parallel SARSA reinforcement learning agents and genetic algorithm

Asghari, Ali; Sohrabi, Mohammad Karim

doi:10.1007/s11227-020-03364-1

Cited by 83 publications

(34 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Given an application comprised of tasks with no interdependence among them, the scheduling challenge consists of assigning each task the most suitable computation resource for its timely execution. However, the proposed evaluation methodology can be applied to workflow scheduling as well in the case of tasks with precedence constraints [12].…”

Section: Task Schedulingmentioning

confidence: 99%

Evaluation of Task Scheduling Algorithms in Heterogeneous Computing Environments

Stan

Băjenaru

Negru

et al. 2021

Sensors

View full text Add to dashboard Cite

This work establishes a set of methodologies to evaluate the performance of any task scheduling policy in heterogeneous computing contexts. We formally state a scheduling model for hybrid edge–cloud computing ecosystems and conduct simulation-based experiments on large workloads. In addition to the conventional cloud datacenters, we consider edge datacenters comprising smartphone and Raspberry Pi edge devices, which are battery powered. We define realistic capacities of the computational resources. Once a schedule is found, the various task demands can or cannot be fulfilled by the resource capacities. We build a scheduling and evaluation framework and measure typical scheduling metrics such as mean waiting time, mean turnaround time, makespan, throughput on the Round-Robin, Shortest Job First, Min-Min and Max-Min scheduling schemes. Our analysis and results show that the state-of-the-art independent task scheduling algorithms suffer from performance degradation in terms of significant task failures and nonoptimal resource utilization of datacenters in heterogeneous edge–cloud mediums in comparison to cloud-only mediums. In particular, for large sets of tasks, due to low battery or limited memory, more than 25% of tasks fail to execute for each scheduling scheme.

show abstract

Section: Task Schedulingmentioning

confidence: 99%

Evaluation of Task Scheduling Algorithms in Heterogeneous Computing Environments

Stan

Băjenaru

Negru

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…In [18], the integration of state action learning and GA models is employed for managing CC resources. Initially, a smart agent schedules the task in the course of the learning procedure by examining the workflows.…”

Section: Fig 1 Overall Resource Provisioning Process In CC Platform [8]mentioning

confidence: 99%

A Novel Fuzzy Clustering with Metaheuristic based Resource Provisioning Technique in Cloud Environment

Al-Masri¹,

Nasir²

2021

FPA

View full text Add to dashboard Cite

Cloud Computing (CC) becomes a commonly available tool to enable quick, on-demand services from a shared pool of configurable computing resources which can be allocated and utilized. Resource provisioning is a major issue in CC environment which ensures guaranteed outcomes on the applications related to CC. This study introduces an efficient fuzzy c-means clustering (FCM) with hybrid grey wolf optimization (GWO) and differential evolution (DE) algorithm, called FCM-GWODE for resource provisioning in cloud environment. The aim of the FCM-GWODE technique is to allocate the resources in such a way that the resource utilization can be accomplished. In addition, the FCM technique with metaheuristics is applied to partition the resources and scalable searching process can be minimized. Moreover, the GWODE algorithm is derived by resolving the local optima issue of the GWO and improve the population diversity using DE. A comprehensive simulation process takes place using CloudSim tool and the results are inspected interms of several evaluation metrics. The simulation results highlighted the supremacy of the FCM-GWODE technique over the other methods.

show abstract

“…The combination of the Sarsa algorithm and other methods has also achieved certain research results, such as the k-means clustering algorithm (k-means) [12] and neural network algorithm [13][14][15].…”

Section: Related Workmentioning

confidence: 99%

Autonomous Driving Path Planning based on Sarsa-Dyna Algorithm

Hassanien¹,

Mononteliza²

2020

APJCRI

View full text Add to dashboard Cite

In autonomous driving path planning, ensuring the computational efficiency and safety of planning is an important issue. The Dyna framework in reinforcement learning can solve the problem of planning efficiency. At the same time, the Sarsa algorithm in reinforcement learning can be effective in guaranteeing the safety of path planning. This paper proposes a path planning algorithm based on Sarsa-Dyna for autonomous driving, which effectively guarantees the efficiency and safety of path planning. The results show that the number of steps planned in advance is proportional to the convergence speed of the reinforcement learning algorithm. The Sarsa-Dyna will be proposed. The analysis of convergence speed and collision times has been done between the proposed Sarsa-Dyna, Q-learning, Sarsa and Dyna-Q algorithm. The proposed Sarsa-Dyna algorithm can reduce the number of collisions effectively, ensure safety during driving, and at the same time ensure convergence speed.

show abstract

Task scheduling, resource provisioning, and load balancing on scientific workflows using parallel SARSA reinforcement learning agents and genetic algorithm

Cited by 83 publications

References 55 publications

Evaluation of Task Scheduling Algorithms in Heterogeneous Computing Environments

Evaluation of Task Scheduling Algorithms in Heterogeneous Computing Environments

A Novel Fuzzy Clustering with Metaheuristic based Resource Provisioning Technique in Cloud Environment

Autonomous Driving Path Planning based on Sarsa-Dyna Algorithm

Contact Info

Product

Resources

About