The Key Technologies Research For Data Service In Mobile Internet

For data services in mobile Internet,on the one hand,efficient data processing models are needed to support data services,on the other hand,a certain amount of computing resources is required to provide guarantee for data security services.This paper mainly focuses on key technology research of personalized data services,data privacy protection and secure data sharing in mobile Internet,to improve the existing methods of data analysis and processing,to realize the fine-grained data access control through effective user privacy protection mechanism,to promote the data sharing in the mobile Internet,and to futher provide technical support for improveing various service applications based on mobile Internet.In the aspect of personalized data services,mobile Internet data is not only large in scale,but also characterized by various patterns,complex connections,and uncertain credibility.Various patterns make data content difficult to be understood,complex connections making data difficult to be effectively identified,and uncertain credibility makes data difficult to determine its authenticity.Such data characteristics make the personalized data services face new challenges.Based on individual differences among user data,this paper proposes three prediction models for smart phone front-end sensor data,backstage server log data and multimedia data in social network,to provid support for personalized service.In particular,the first model puts forward a new method of using intelligent mobile phone application response time difference to predict gender information.By using the log data of smart phone application access to cyber source,the second model predicts the user's gender,age and the type of smart phone.It matches the entries in the smart phone application log with its subject categories,and proposes an optimization method for prediction results based on the Bias classifier.The third model proposes a new way to predict user gender,terminal type,activity and influence by using social pictures released by users.In the aspect of data privacy protection,proxy re-encryption(PRE)enables a semi-trusted proxy to delegate the decryption right by re-encrypting the ciphertext under the delegator's public key to an encryption under the public key of delegatee.Fueled by the translation ability,PRE is regarded as a promising candidate to secure data sharing in a cloud environment.However,the security of the PRE will be totally destroyed in case the secret key of the delegator or the delegatee has been exposed.To ease the destructive result of key exposure,we propose a key-insulated proxy re-encryption(KIPRE)scheme in this thesis.In the proposed scheme,the lifetime of the secret key associated with the user has been divided into several periods.In each time period,the user can interact with his/her physically-secure but computation-limited helper to update his/her temporary secret key.On the contrary,the public keys of the users remained unchanged during the whole lifetime of the system.We then apply our KIPRE scheme to construct a practical solution to the problem of sharing sensitive information in public clouds with resilience to the key exposure.To ensure the security of the outsourced data in the cloud environment,sensitive data must be encrypted before being outsourced to the server.The traditional PEKS(Public Encryption Keyword Search)enables users to search data by using keywords in the condition of encryption,however,exising schemes either tolerate the huge pairing-computation,or lack of the support of fuzzy keyword searc.In this thesis,a pairing-free PKES scheme is proposed,which features the following good aspects: First,our scheme not only supports accurate keyword search encryption but also supports the search when the keywords input have any spelling mistakes or format inconsistencies,which significantly improved the availability of the system.Second,the proposed scheme is constructed on the El Gamal encryption instead of the bilinear-pairing encryption,which greatly improves the computational efficiency.The performance evaluation and the security analysis demonstrate that both data sharing scheme are efficient and feasible.In the aspect of secure data sharing,to ensure the outsourced data in the untrusted cloud server can only be accessed by the lightweight mobile device,attribute-based encryption(ABE)is commonly regarded as the essential technology to design security solutions for the cloud due to its “one-to-many” encryption property.By combing the public key encryption with equality test and secret key blinding technique,a cloud-assisted secure data sharing scheme has been proposed for the mobile Internet based on the attribute-based encryption.On the one hand,the untrusted cloud server can be delegated to perform the search functionality on the outsourced ABE-type ciphertext by using the idea of public key encryption with equality test.Furthermore,the ciphertext can be searched even the attributes associated with the keyword doesnot match the access policy of the ciphertext.On the other hand,the heavy computational overhead in the decryption can also be offloaded to the untrusted cloud server,which makes the proposed scheme is especially suitable for the mobile device with limited resources.The proposed scheme is formally described with finite state machine model.The performance evaluation and the security analysis demonstrate the suggested data sharing scheme is efficient and feasible.