Abstract
The decentralized many-to-many negotiation for resource allocation in Cloud and multi-agent systems presents numerous challenges, including ones related to the buyer strategy which is the focus of the present paper. Current approaches deriving required resources each bid must ask for aren't in all market cases an optimal choice. For this reason, we have proposed a hybrid negotiation strategy consisting of a combination of two modes of negotiation strategies that generates required resources of each bid in parallel, the first one is an existent fixed negotiation strategy and the second one is a learning selection strategy over the buyer's agreement space. Moreover, acting dynamically in the market place by adjusting appropriately the buyer's resource provisioning times and calling for proposal to hand over contracted resources in order to break some deadlocks involving buyers' tasks has been shown via simulation results to achieve better performances both in terms of social welfare and buyer utility.