Affiliations: [a] Department of Computer Science, University of Badji Mokhtar, P.O. Box 12, 23000 Annaba, Algeria | [b] INSA de Rouen, 76800 Saint-Etienne du Rouvray, France
Abstract: In the automated many-to-many negotiation model for leasing resource contract under Cloud computing environment, the main concern will be over finding a good approximation approach to this type of negotiation problem, since an optimal solution to that problem would be intractable due to the inherent complexity of the leasing resource negotiation along with the Cloud’s specifications. In order to get that good approximation, decentralized negotiation, leveled commitment contracts as well as lightweight agent strategies had been proposed, resulting to a fast and well-approximate negotiation decisions. To this end, we had conceived a “composite-greedy” algorithm for the seller’s strategy to address a variant of WDP which is the seller’s main problem under such settings. The composite greedy seller strategy will be a two-step strategy: the first one applies a classification according to the offers’ revenue and the second one according to the offers’ densities, and thereafter the seller will pick the one that maximizes the most his expected utility. Experimental results show the effectiveness of our approach compared to some related ones over different performance measures.