带隐私保护的群智感知任务分配机制

Privacy-preserving Allocation Mechanisms for Crowdsensing Tasks

存在多任务发布者的公共群智感知平台存在不可信问题,但现有机制均无法在任务分配过程中保护任务发布者的隐私信息,影响了群智感知系统的应用和发展.针对该问题,文章研究了如何在保证分配机制正常运行的前提下,同时保护任务发布者的预算、收益函数以及用户的敏感信息不泄露.所设计的机制引入了半可信第三方,通过动态IP交互、同态加密扰乱、置换以及数字签名等技术,保证了平台和半可信第三方均无法获得相关价格隐私,且无法将用户与其所提交数据相关联,从而进一步预防泄露用户数据的潜在隐私.所设计的机制可以在任务匹配和支付过程中保护隐私,且能实现所有任务发布者收益最大化.通过仿真实验结果验证了机制的有效性和可行性.

The crowdsensing platform may be untrustworthy when multiple task requesters share a common platform. However,none of the existing mechanism can protect the privacy of task requesters in the procedure of task assignment,which will affect the application and development of the crowdsensing systems. To solve this problem,we propose a crowdsensing task allocation mechanism,which can protect the budget,profit function of task requesters and the sensitive information of users simultaneously. The proposed privacypreserving task allocation mechanism introduces the semi-trusted third parties. Since the sensing data of users may leak their privacy,we need to prevent the platform or the requester from associating the users with their submitted sensing data. To achieve this,the proposed mechanism lets users communicate with the platform or requesters by using dynamic IPs,and uses RSA digital signature technology to finish the payment between users and requesters. Further to protect the bid privacy of users and the budget privacy of task requesters,homomorphic encryption operation,data disturbance and permutation are used in process of crowdsensing task allocation. We prove that the proposed mechanism can maximize the total revenue of all task requesters,and can protect the sensitive information of users and budgets of task requesters in the task allocation and payment process. Finally,the effectiveness and feasibility of the proposed task allocation mechanism are verified by the simulation results.