Dynamicevaluation Mechanism Of Privacy Protection On Dynamic Data Targeted At Multi-Tenant Application

With the development of information technology and the popularity of cloud services, more and more enterprises or local organizations outsourced large amounts of data and complex management to cloud service providers, reducing investment costs of tenant infrastructure, subsequent upgrades, maintenance, management, etc. At the same time, due to the outsourcing of data in the cloud is not directly controlled and managed by tenants. It will face the threat of privacy leakage external attacks and management cloud service providers. Faced with these threats, researchers have proposed a number of privacy-protection technologies. The current mainstream technology, including data distortion, data encryption, partition confusion, etc. Data distortion technique is to make sensitive data distort while maintaining some data and attributes the same, but mainly used in the field of statistics. Data encryption will protect sensitive data by using encryption technology, but the larger process of decryption and encryption overhead. The technology of partition-confusion partitions the data and confuses relationships between attributes to protect user privacy according to the data privacy constraints under circumstance of ensuring the distortion and application performance of data. Such as{name, age, address, salary} is a privacy constraints that can uniquely identify a user address information. The Privacy Constraints properties are divided into different data blocks by security relationship between the third-party storage encryption, so an attacker cannot determine a complete record, and thus protecting the privacy of data tenants.The research of privacy protection technology based on partitions is mainly focused on user’s privacy constraints to sub partitions, while the protection of privacy of association rules and data distribution under continuous dynamic changes of the data set has not been investigated. The tenant’s data often has some hidden inherent relationships, and along with the continuous update of data set, the data relationship of the attributes and data distribution will also change. Especially with the development of data mining technologies and related applications, some of the attributes that appear to be not privacy sensitive can also be inferred, and because the local data may not be balanced, but also very vulnerable to guessing attacks, thereby leaking privacy. Therefore, how to detect these tenants behavior o and abnormal data distribution that leaks tenant’s privacy and the properties of their privacy for further protection has become particularly important.In response to these problems, we seek a protection mechanism for the evaluation, it evaluates the degree of privacy leakage when data changes dynamically, then adjust the data protection for tenants.1. The privacy leakage of association rules:This article first dynamically testing correlation of tenant block data determine whether disclosure of private information of tenants, and then based on the dimension of properties that leaks privacy make a selection is to add noise or re-blocking, and it need to optimize application according to tenants operating practices to ensure performance requirements. In this paper, the case of privacy leak of tenant’s data divided into association rules leakage and uneven distribution of data leakage based on the analysis of the block confuse privacy protection technology, Then two cases were evaluated dynamically and measure the degree of protection, and finally adjusted the user’s set of data according to the degree of privacy leakage of tenant’s data to ensure tenant’s data will not be disclosed2. The privacy leakage of imbalanced data distribution:The thesis proposes a kind of leakage degree evaluation algorithm based on privacy leakage situation possibly arising in data combined privacy protection mechanism based on partitioning, and finds Imbalance data and the relevance among possibly data partition, and then makes real-time evaluation of the degree of privacy leakage from the perspective of internal attack to external attack based on information entropy.Through the above research, this paper presents a multi-tenant application for privacy protection dynamic evaluation mechanism, and by experimental verification of the feasibility and effectiveness of the mechanism algorithm from multiple angles.