Dynamic clustering of tasks and DCS for multiple task allocation

Abstract

This paper describes how the allocation of stream of tasks, with minimum knowledge, is possible in a distributed computing system. In literature, almost all the task allocation models in a distributed computing system require a priori knowledge of tasks execution time, communication time etc. on the processing nodes. Since the task assignment is not known in advance, this time is difficult to estimate. A cluster-based dynamic allocation scheme is proposed for both the distributed computing system and the tasks that eliminate the execution time requirement. Further, as opposed to a single task, multiple tasks are considered for allocation by the model. For both the task clustering and processor clustering a fuzzy function is used. Clustering and assignment process is used dynamically as it suits the stochastic stream of incoming tasks. Experimental validate the efficacy of the proposed model. © 2005 Springer Science+Business Media, Inc.