完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.author | HSU, CC | en_US |
dc.contributor.author | LEE, PJ | en_US |
dc.date.accessioned | 2014-12-08T15:04:17Z | - |
dc.date.available | 2014-12-08T15:04:17Z | - |
dc.date.issued | 1993-11-01 | en_US |
dc.identifier.issn | 0020-0255 | en_US |
dc.identifier.uri | http://hdl.handle.net/11536/2803 | - |
dc.description.abstract | A very general task assignment scheme for a distributed computing system is presented in this paper. It takes execution time, communication time, and idle time into consideration. All possible task assignments are formed as a state space tree where a path from the root node of the tree to a node represents a partial or total task assignment. We propose an approach called the critical path overestimate for achieving a suboptimal task assignment and pruning most of the nodes of the tree. Combining the above approach with a divide-and-conquer method, we can further reduce the complexity of the problem, and therefore attain a satisfactory solution. Results from a wide range of experiments reveal that the proposed approaches perform well because of their giving close approximation to the actual cost and saving a high percentage of node generations. | en_US |
dc.language.iso | en_US | en_US |
dc.title | DISTRIBUTED TASK ASSIGNMENT USING CRITICAL PATH ESTIMATE | en_US |
dc.type | Article | en_US |
dc.identifier.journal | INFORMATION SCIENCES | en_US |
dc.citation.volume | 74 | en_US |
dc.citation.issue | 3 | en_US |
dc.citation.spage | 191 | en_US |
dc.citation.epage | 212 | en_US |
dc.contributor.department | 資訊科學與工程研究所 | zh_TW |
dc.contributor.department | Institute of Computer Science and Engineering | en_US |
dc.identifier.wosnumber | WOS:A1993MC65900001 | - |
dc.citation.woscount | 1 | - |
顯示於類別: | 期刊論文 |