DocumentCode :
499173
Title :
Earliest Starting and Finishing time Duplication-based algorithm
Author :
Hosseinzadeh, Mahsa ; Shahhoseini, Hadi Shahriar
Author_Institution :
Electr. Eng. Dept., Iran Univ. of Sci. & Technol., Tehran, Iran
Volume :
41
fYear :
2009
fDate :
13-16 July 2009
Firstpage :
49
Lastpage :
56
Abstract :
Optimal task scheduling of a direct acyclic graph (DAG) onto distributed environments is a NP-hard problem. In this paper, we propose a new scheduling algorithm called earliest starting and finishing time duplication based (ESFD). ESFD has three phases: priority processing, task scheduling and task duplication. ESFD considers all parameters related to the task and its immediate predecessors to assigning a task. After each schedule ESFD will be updated its first phase, so ESFD is a dynamic algorithm. Experimental results on random graphs and real application graphs show that the NSL and speedups generated by the ESFD are better than those generated by the well-known HCPT, HEFT and LHCNF algorithms.
Keywords :
computational complexity; directed graphs; distributed processing; dynamic programming; scheduling; DAG; ESFD; NP-hard problem; direct acyclic graph; dynamic algorithm; earliest starting-and-finishing time duplication-based algorithm; heterogeneous distributed computing system; optimal task scheduling algorithm; priority processing; Clustering algorithms; Communications technology; Computational modeling; Distributed computing; Dynamic scheduling; Finishing; Heuristic algorithms; NP-hard problem; Processor scheduling; Scheduling algorithm; DAG; Duplication; Heterogeneous systems; List scheduling; Task scheduling;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Performance Evaluation of Computer & Telecommunication Systems, 2009. SPECTS 2009. International Symposium on
Conference_Location :
Istanbul
Print_ISBN :
978-1-4244-4165-5
Electronic_ISBN :
978-1-56555-328-6
Type :
conf
Filename :
5224144
Link To Document :
بازگشت