The Privacy Preserving Of Aggregation In New Network Environment

With the rapid development of network information technology,the creation of new net-work platform such as smart grid,social network,internet of things,participatory sensing and cloud computing,data collection and sharing become more and more convenient.But at the same time,people are increasingly concerned about the possible disclosure of data privacy.Today,the data are considered to be the most valuable resource in the world.Data mining tech-niques,including data aggregation,are experiencing rapid development due to the need for data processing.This also brings more and more challenges to the privacy preserving.To research the privacy preserving of data aggregation in these new networks has not only the important value in theory but also is necessary in the real applications.And,it is very important for the national information construction.This thesis focuses on the researches of privacy preserving of data aggregation in smart grid and multi-domain wireless network.The followings are the main works and contributions of this thesis:?1?Through the deep research on differential privacy guarantees,this thesis proposes the binomial distribution mechanism and Poisson mechanism which fit for the distributed noise addition.Differential privacy attack is a new attack which could threaten all existing secure aggrega-tion protocols;only those protocols with differential privacy could survive from this attack.The differential privacy are guaranteed by adding random noises on the real answers.After adding the noises,the statistical distribution distance of two data sets D₁and D₂which have only one different record does not excess than e^?.So,even if one participant has removed his data from the data set,the output result has no obvious changes.Most of the existing differential private mechanism must depend on a trusted data aggregator which adds the noises on the data before the statistics are released.But,in many distributed network settings,there is no such trusted aggregator.So,we propose the mechanism for the distributed noise addition.Under the inter-active frame,considering with the characters of smart grid and multi-domain wireless network,we propose the binomial distribution mechanism and poisson mechanism basing on the origi-nal Laplace mechanism.This makes it possible to convert differential privacy that can only be implemented by a trusted center into a privacy protection by the users community themselves,thus making the differential privacy theory more broadly applicable to a network environment that has no trusted center.?2?Through the deep research on the privacy preserving of the data aggregation for smart grid,this thesis firstly identify and formalize a new attack and proposes a privacy-preserving aggregation protocols to resist the attack.The existing privacy preserving aggregation protocols proposed for smart grid only con-sider that for the untrusted data aggregator,the real meter data of the user is not disclosed in the aggregation.But we take into account the possible malicious data mining on the released result.Because individual activities make a difference in power consumption in a certain period of time,this distinction corresponds to a differential privacy attack on data mining in the database.Therefore,we first propose a new type of attack model,coined as the human-factor-aware dif-ferential aggregation attack,abbreviated as HDA attack.We give a formal definition on it and propose two novel protocols to achieve privacy-preserving smart metering data aggregation and resist the HDA attack.These protocols take into account the availability and security of the data,and support efficient data aggregation for time-series metering data without leaking the in-dividual value by using the method of the distributed noise addition and encryption technology.We give a formal proof to the security of the proposed protocols and a detailed performance evaluation of the proposed protocols by the simulation.The performance and utility analysis shows that our protocols are simple,efficient and practical.?3?The privacy preserving of the data aggregation in the multi-domain wireless sensor networks is studied,we present a novel hybrid cloud base privacy preserving outsourced data aggregation framework and two efficient and secure privacy-preserving aggregation protocols under this framework.With the increasing popularity of mobile communication networks and wireless networks,multi-domain sensor networks formed by participatory sensing are gaining more and more at-tention.Enabling privacy preserving outsourced data aggregation is regarded as an important issue for multi-domain wireless networks.As the calculation and storage of the sensor are lim-ited,we can combine the cloud computing platform to complete the operation of data aggrega-tion.Because the existing privacy preserving architecture is not suitable for the development of multi-domain wireless networks and cloud computing technologies,we have proposed a new cloud-based privacy preserving architecture?CPPA?for the first time,which combines the multi-domain wireless network and the cloud computing,in one model to achieve the data storage and aggregation operations at the same time.Individuals only need to encrypt the data before up-loading to the cloud which will be able to flexibly respond to a variety of different needs of the polymerization requirements.Secondly,according to the separability and non-separability of information content,we propose two novel protocols,including the pro-active privacy preserving data aggrega-tion?PPPA?and reactive privacy preserving data aggregation?RPPA?schemes,which are based on the idea of secret sharing.The pro-active scheme allows the user to pro-actively split their data to multiple storage clouds to avoid data leaking while the reactive scheme allows the users to store their encrypted data in storage cloud and aggregator to finish the data aggregation based on the encrypted data.Under the CPPA architecture,these two protocols support privacy pre-serving aggregation.Moreover,based on the PPPA and RPPA,we further propose an advanced protocol which can resist the malicious data mining?MDM?attack.The detailed performance simulations are given to show that the proposed schemes have the enough security,effectiveness and efficiency.In summary,this paper systematically studies the privacy protection of the aggregation op-eration under the new network model.The proposed security protocols not only have important theoretical significance,but also have practical application value in engineering.