Differential Private Data Collection and Analysis Based on Randomized Multiple Dummies for Untrusted Mobile Crowdsensing
Differential Private Data Collection and Analysis Based on Randomized Multiple Dummies for Untrusted Mobile Crowdsensing
复制标题
DOI：
10.1109/tifs.2016.2632069
复制
发表时间：
2017-04
影响因子：
6.8
通讯作者：
Y. Sei;Akihiko Ohsuga
中科院分区：
文献类型：
--
作者：
Y. Sei;Akihiko Ohsuga
从移动的电话用户收集环境信息的移动的人群感知越来越受欢迎。这些数据可以被公司用于营销调查或决策。然而，从其他用户收集感测数据可能会侵犯他们的隐私。此外，众测的数据聚合器和/或参与者可以是不可信实体。最近的研究提出了匿名数据收集的随机应答方案。这种数据收集可以对用户的感知数据进行统计分析，而无需关于其他用户的感知结果的精确信息。然而，传统的随机响应方案及其扩展需要大量的样本来实现适当的估计。在本文中，我们提出了一个新的匿名数据收集方案，可以更准确地估计数据分布。使用模拟与合成和真实的数据集，我们证明了我们所提出的方法可以减少均方误差和JS分歧超过85%，与其他现有的研究相比。
Mobile crowdsensing, which collects environmental information from mobile phone users, is growing in popularity. These data can be used by companies for marketing surveys or decision making. However, collecting sensing data from other users may violate their privacy. Moreover, the data aggregator and/or the participants of crowdsensing may be untrusted entities. Recent studies have proposed randomized response schemes for anonymized data collection. This kind of data collection can analyze the sensing data of users statistically without precise information about other users’ sensing results. However, traditional randomized response schemes and their extensions require a large number of samples to achieve proper estimation. In this paper, we propose a new anonymized data-collection scheme that can estimate data distributions more accurately. Using simulations with synthetic and real datasets, we prove that our proposed method can reduce the mean squared error and the JS divergence by more than 85% as compared with other existing studies.