隐私保护众包软件测试激励机制设计

Incentive mechanism design for privacy protection in crowdsourced software testing

针对众包软件测试中存在的激励攻击以及测试工人的隐私保护问题,该文提出了一种基于拍卖机制和差分隐私的隐私保护众包软件测试激励机制。拍卖机制能够保证工人报价的真实性,防止激励攻击。差分隐私机制为工人分配被选中的概率,并按照概率分布选择工人,实现测试工人的成本隐私保护。通过理论分析,证明了该文提出的激励机制能够满足计算有效性、真实性、个体理性、差分隐私和近似招募成本最小。试验表明该文提出的激励机制仅以平均总成本增加27.31%的代价保证了工人的测试成本隐私。

Aiming at the incentive attack and privacy protection of workers in crowdsourced software testing,this paper proposes an incentive mechanism for privacy protection in crowdsourced software testing based on auction and differential privacy.The auction mechanism can ensure the authenticity of the quotations of workers and prevent the incentive attack.The differential privacy mechanism assigns the probability of workers being selected,and selects workers according to the probability distribution,which can realize the privacy protection of workers’testing cost.Through theoretical analysis,it is proved that the designed incentive mechanism can satisfy computational validity,authenticity,individual rationality,differential privacy and approximate minimum recruitment cost.Experiments show that the proposed incentive mechanism sacrifices an average 27.31%of the total cost to ensure the privacy of workers.