A Novel Grid Resource Scheduling Model Based on Extended Second Price Sealed Auction

摘要

In resource-limited environment, grid users compete for limited resources, and how to guarantee tasks¡¯ victorious probabilities is one of the most primary issues that a resource scheduling model cares. In order to solve grid resources scheduling problems, a novel model, namely ESPSA (Extended Second Price Sealed Auction), is proposed. The ESPSA model introduces an analyst entity, and designs analyst¡¯s prediction algorithm based on Hidden Markov Model (HMM). In ESPSA model, grid resources are sold through second price sealed auction. Moreover, to achieve high victorious probabilities, the user brokers who are qualified to participate in the auctions will predict other players¡¯ bids and then carry out the most beneficial bids. The ESPSA model is simulated based on GridSim toolkit. Simulation results show that the ESPSA model assures a higher victorious probability and superior to other traditional algorithms. Moreover, we analyze the existence of Nash equilibrium based on simulation results.