Load balancing on multi-computer systems다중컴퓨터에서 일의 균등분배에 관한 연구

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 575
  • Download : 0
In a multi-computer systems, a job is usually decomposed into several cooperating modules which are then assigned to a set of processors in the system to exploit the inherent parallelism in job execution and to make use of computational capabilities and resources of the system efficiently. In this thesis, we investigated the problem of assigning tasks onto multi-computers. The problem involves the development of task allocation models for the multi-computer system. The model must allocate tasks among processors to archive the following goals: - To accommodate specification of a large number of constraints to facilitate a variety of engineering application requirements. - To balance the utilization of individual processors in the distributed computing system. and - To minimize interprocessor communication cost. In this thesis, we first deal with static task assignment problem for the host-satellite multi-computer system. The assignment problem, which is the problem of assigning the task modules of a task to the processors of a multi-computer system is known to be NP-complete in its most general cases. Previous work on this problem has shown in tractable for a system of two processors[Stone77]. We propose a graph theoretic approach which finds optimal solutions for the general multiple-satellite system. Our work is an extension of the Stone``s two-processor approach to the case for the multiple-satellite system. The problem is solved by applying the well-known network flow algorithm in the related two-terminal network graphs. Therefore, the task assignment problem in a homogeneous satellite system can be solved in time no worse than $O(NM^3)$ by applying the network flow algorithm, where N and M are the number of satellites and the number of modules, respectively. Secondly, the task allocation problem for the heterogeneous multi-computer system is concerned. In general, we are considering a heterogeneous processor system in which each processor may have different...
Advisors
Park, Kyu-Horesearcher박규호researcher
Description
한국과학기술원 : 전기및전자공학과,
Publisher
한국과학기술원
Issue Date
1995
Identifier
101741/325007 / 000885204
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 전기및전자공학과, 1995.8, [ vi, 110 p. ]

Keywords

Heuristic; Load Balancing; Task Allocation; Cost Function; 비용함수; 휴리스틱; 균등분배; 일의 할당

URI
http://hdl.handle.net/10203/36299
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=101741&flag=dissertation
Appears in Collection
EE-Theses_Ph.D.(박사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0