Incentive Mechanism Study Of Data Collection And Storage Systems In Crowdsourcing And Cloud Computing

The development of digital world,the advance of cloud storage technology,the explosive growth of big data,along with the needs of user privacy protection and data security have given birth to a series of secure and efficient data collection and storage technologies.Blockchain-based decentralized crowdsourcing data collection systems remove the dependence on the credibility and reliability of a centralized platform.They apply consensus mechanisms for enabling miners to release and arrange tasks,calculate the rewards of system nodes and achieve fair payment in a distributed way.Encrypted cloud data deduplication employs a series of cryptographic algorithms to check the duplication of encrypted data,verify the ownership of data users,and control data access and management of eligible users.However,it is still under discussion whether the above systems can be deployed in practice and accepted by the public.We discovered the following problems of the incentive study on existing data collection and storage systems in crowdsourcing and cloud computing.First,there is still a lack of systematic analysis of their applications in practice.Second,existing incentive mechanisms primarily assume the system entity that dominates the system will actively participate;therefore,the incentive of this entity is ignored in incentive mechanism design.Third,the incomplete consideration of motivated system entities further causes the problem of incomplete goals of incentive mechanism design.However,the rationality and the profit maximization objective of system entities in practical systems and the preferences of the incentive mechanism designer require a feasible incentive mechanism to achieve multiple objectives for collaborative motivation.We summarize the practical adoption problems of data collection and storage systems in crowdsourcing and cloud computing as the problem that whether the system entities are willing to participate and cooperate by behaving honestly,and put forward a series of incentive mechanisms under the application scenarios of decentralized crowdsourcing systems and encrypted cloud data deduplication.Our contributions are specified as follows:(1)Aiming at the problems of incomplete consideration of motivated system entities and unilateral goal of incentive mechanism design in current decentralized crowdsourcing data collection systems,this dissertation proposes a grade-driven auction-based incentive mechanism with multiple objectives.We first design a grade-based task sorting algorithm to prioritize services for heterogeneous crowdsourcers,based on which,we propose a hierarchical premium-based worker selection method for balancing social grade maximization and social cost minimization goals by considering the capability limitations of workers and the task requirements of crowdsourcers.Finally,we design a fair reward sharing method to stimulate high-grade miners on the premise of guaranteeing the profits of crowdsourcers.(2)In view of the above three problems in server-controlled encrypted cloud data deduplication,we first put forward an economic model for this scenario and model the interactions between a cloud storage provider(CSP)and data holders as a non-cooperative game,where their strategy is whether to participate or not.We further analyze the feasibility conditions of a discount-based mechanism under the incentive requirements of individual rationality,incentive compatibility,profitability and robustness.In order to satisfy the incentive compatibility of CSP,we propose a bounded discount-based incentive mechanism,which improves the participation willingness of data holders and CSP.(3)In order to solve the same three problems in client-controlled encrypted cloud data deduplication,based on the research in Contribution 2,we consider how to eliminate freeriding behaviors of data holders and avoid privacy leakage.We model the interactions of a data holder and a data owner as a non-cooperative game,where their strategy is whether to cooperate by behaving cooperatively or not.Through equilibrium analysis,we discover the free-riding problem in a unified discount mode and privacy leakage problem in an individualized discount mode.We further provide an improved unified discount-based incentive mechanism and an individualized discount-based incentive mechanism with privacy protection,which realize the goals of individual rationality,incentive compatibility and profitability.(4)Under hybrid controlled encrypted cloud data deduplication,the interaction among a CSP,a data owner and data holders is modeled as a multi-stage Stackelberg game with perfect information.Based on Contribution 2 and Contribution 3,we explore the practical feasibility of discount-based incentive mechanisms by analyzing the existence of game equilibrium with a reverse induction method.Namely,we solve the Stackelberg game reversely in the order of a holder participation game,an owner online game and a CSP pricing game.We also put forward a gradient-based strategy searching algorithm for determining the near-optimal strategy for the CSP in a limited time,which overcomes the challenge that it is difficult for a practical system entity to accurately calculate the equilibrium state.