| 摘要: |
| 针对移动群智感知中参与者数据的真值和隐私保护问题,提出了一种高效可验证的安全真值发现方法EVSTD,通过安全迭代更新参与者权值和评估对象真值,从而得到对象的真实数据。EVSTD中,参与者利用本地随机数和协商随机数对敏感数据进行双掩码数据扰动,使得EVSTD不仅能够保证敏感数据的隐私性,且解决了参与者因延迟发送感知数据而导致的敏感数据泄露问题。同时,EVSTD利用秘密共享协议解决了参与者掉线或失效的问题,且通过动态选择L邻居节点策略让参与者只与其关联邻居进行通信从而大大降低了参与者的计算和通信开销。此外,参与者通过计算敏感数据的同态哈希值以用于数据的验证并上传给服务器,服务器对敏感数据进行聚合和对验证信息进行乘积,并将计算结果发送给参与者,参与者再对聚合结果和证明信息进行验证,验证通过则说明聚合结果正确,进一步保证了真值发现结果的可信性,防止服务器对参与者的敏感数据进行篡改,保证了聚合结果的真实性。实验结果显示所提方法在保证数据隐私的同时获得真实可靠的数据信息,且能够有效的防止服务器篡改数据和共谋攻击。 |
| 关键词: 移动群智感知 真值发现 数据隐私 验证 双掩码 |
| DOI：10.19363/J.cnki.cn10-1380/tn.2024.03.09 |
| Received:May 03, 2022Revised:September 18, 2022 |
| 基金项目:本课题得到国家自然科学基金项目(No.62272006,No.61972438)、安徽省重点研究与开发计划项目(No.2022a05020049)、安徽省自然科学基金项目(No.2108085MF219)、安徽省教育厅高校自然科学研究重点项目(No.2023AH052695)资助。 |
|
| An Efficient and Verifiable Secure Truth Discovery in Mobile Crowdsensing |
| WANG Taochun,ZHANG Chenlu,CAI Songjian,CHEN Fulong,SHEN Huimin,XIE Dong |
| School of Computer and Information, Anhui Normal University, Wuhu 241002, China;Anhui Engineering Research Centers of Medical Big Data Intelligent System, Anhui Normal University, Wuhu 241002, China |
| Abstract: |
| Aiming at the privacy protection of participant’s truth data in mobile crowdsensing, we proposed an efficient and verifiable security truth discovery method, named EVSTD. By the security iterations, the participant’s weight value and the truth of the evaluated object are updated, so we can obtain the truth of the object. In EVSTD, participants use local seed to generate a local random number, and negotiate seed with associated neighbors to generate a negotiated random number by the key agreement protocol. Participants use local random number and negotiated random number to disturb sensitive data with double masks, which can not only ensure the privacy of sensitive data, but also solve the problem of sensitive data leakage caused by delayed sending of perceived data by participants. At the same time, EVSTD uses secret sharing protocol to solve the problem of disconnection or invalidation of participants, and the strategy of select L-neighbor node dynamically to let participants only communicate with their associated neighbors, thus greatly reducing the computational and communication costs of participants. At the same time, when the participant generates disturbed data, it calculates the homomorphic hash of sensitive data and the data used for verification and uploads them to the server. The server calculates the aggregation result and its proof according to the sensitive data and verification information, and finally sends them to participants. The participant verifies the aggregation results and proof data sent by the server according to the existing verification information. If the verification passed, the aggregation result will be correct, which further guarantees the credibility of the truth discovery results, so as to solve the problem that the cloud server may tamper with the participant's sensitive data, and ensure the reality of the aggregation results in mobile crowdsensing perception. The experimental results show that the proposed method can identify the true and reliable data information while protecting data privacy, and can prevent servers from tampering with data and conspiracy attacks. |
| Key words: mobile crowd sensing truth discovery data privacy verification double mask |
