Abstract: Considering the problem of unauthorized access and malicious attacks in multi-tenant application, a Multi-Tenant Privacy Protection Model（MTPPM） is presented using the lossless decomposition theory of relation database. The model can classify the least attributes according to attribute privacy constraint, and process the data with different anonymization approaches, and then achieve the balanced distribution through the disturbance of the data. It could also solve the problem of private information retrieval to protect user privacy by applying reconstruction method on the data. The theoretical analysis and experimental results show that the new model can provide efficient data privacy protection, and consider the requirements for the user under the trusted third party.

Key words: multi tenancy, privacy protection, cloud computing, lossless decomposition

摘要： 针对多租户应用中存在的越权访问和联合攻击问题，利用关系数据库理论的无损分解思想，提出了一种多租户环境下的隐私保护模型。该模型根据属性隐私约束进行最少属性分解，并采取不同的匿名化处理方法，进而通过数据扰动实现数据的平衡分布，同时重构调整后的数据，以实现数据库访问时的隐私保护。通过分析和实验，算法能够较好地保护用户隐私数据，在可信第三方环境下兼顾了租户的自定制需求。

关键词: 多租户, 隐私保护, 云计算, 无损分解