Skip to Main Content
The aim of this paper has been to introduce a novel market based algorithm for grid resource allocation and resolve some incompetence with current mechanisms. Grid resource allocation could be considered as a double auction in which grid resource manager acts as an auctioneer and jobs and resource owners act as buyers (resource consumers) and sellers. In our approach resource allocation is based on activity history of each participant in the auction. Bids (Asks) in the auctions are calculated in each round, based on proportion of former bids (asks) fluctuation and prior activity of each buyer(seller). Practical evaluation results demonstrate that market equilibrium in our mechanism is improved in comparison with other mechanisms. Also essence of learning features in proposed mechanism increases competition incentive in participants which do selfishly in every round of an auction to maximize their utility.