A genetic algorithm-based approach to create a safe and profitable marketplace for cloud customers

Designing effective shill bidding detection and prevention mechanisms in a cloud auction house is one of the main challenges of the cloud market. In this paper, one mechanism for shill bidding detection and one for its prevention are focused on. Two objectives are considered in designing these mechanisms: (1) increase the accuracy of shill bidding detection mechanism, and (2) decrease fraud activities of shill bidders while increasing the profit of honest bidders. The accuracy of a shill bidding detection mechanism can be improved by combining results of run-time monitoring of bidding behavior in running an auction and results of bidding behavior obtained from past auctions. Thus, a new hybrid shill detection mechanism is proposed. Also, our idea in designing of shill bidding prevention mechanism is shaped based on the fact that shill bidders continue their fraudulent behaviors only when they are in trading spaces that are created by sellers who have colluded with them. To do this, a genetic algorithm (GA)-based approach is developed to create appropriate trading spaces for honest bidders aiming at minimizing suspicious activities as well as maximizing trading opportunities. Consequently, honest bidders are hosted by more profitable and healthier trading spaces in which the probability of meeting shill bidders and fraud sellers is decreased dramatically. The proposed ideas are supported by a multi-agent auction system. Simulation results prove the success of the designed auction system.