Open Access Open Access  Restricted Access Subscription Access
Cover Image

Multi-Objective PSO Scheduling with Inertia Strategies for Scientific Workload Scheduling in Cloud Environment

Somaprathibha D. A, Latha B, Sumathi G


Cloud scheduling is an important and challenging task in the recent years. In such a scenario, the algorithms used for cloud scheduling must consider the distributed computing nature of cloud and must take care of storage, computing and networking resources and the virtualization features of cloud. The existing scheduling algorithms for process scheduling are suitable for task scheduling in the centralized and distributed operating systems with predefined size of memory, number of processors and platform. On the other hand, the existing scheduling techniques with time sharing basis and limited resources which are working on both deadline based systems and conventional operating systems do not consider the scalability issue effectively. However, cloud computing is more suitable for running complex and processor intensive applications including scientific computation which need to execute a number of tasks in parallel to reduce the execution time. Since, optimal scheduling algorithms are not proposed for cloud scheduling. In this paper, we propose a new Weighted Multi-Objective Particle Swarm Optimization (W-MOPSO) algorithm for scheduling scientific workload in the cloud environment at the Virtual Machine level which is used Multi-objective particle swarm optimization as materials and methods. The inertia weight value affects the convergence, explorations and exploitation features of the PSO algorithm. We extend our work by carrying out performance analysis of W-MOPSO with four inertia weight calculation strategies namely GLIW, TVIW, CIW and RIW for scientific workloads. The main aim of this work is to improve the task scheduling performance in cloud environment. The experimental results show the performance of the proposed algorithm and also compared with existing approaches.

Full Text:



  • There are currently no refbacks.