Combinatorial Double Auction-Based Service Allocation Using an Extended NSGA-III in Clouds

Abstract

In this paper, a service allocation model based on an improved combination double auction (ICDASA) is presented. We have presented a new reputation scheme and integrated it into ICDASA in order to suppress dishonest participants in the cloud market. The proposed model considers five objectives, namely maximizing social welfare, total providers' expectation reputation obtained by consumers, total consumers' expectation reputation obtained by providers, total resource utilization and total task completion percentage respectively. To solve the service allocation problem, which is also a many-objective problem, with some complex constraints, we introduce the constraint handling rules into an extended NSGA-III to refine infeasible solutions and the best compromise solution selection method based on fuzzy set theory. Experiment results have demonstrated that the presented cloud service allocation model generates high social welfare, increases the resource utilization and task completion percentage, and improves the providers'/consumers' expectation reputation obtained by consumers/providers.