Research On Distributed Online Data Analysis Engine For Privacy Protection

With the increase of computing power of mobile devices and the improvement of user privacy awareness,online data analysis has received increasing attention.Unlike traditional data analysis,online data analysis directly performs data analysis on the device side,thereby protecting privacy and security by avoiding the upload of raw data.However,research has found that the current mainstream paradigm of online data analysis still has shortcomings.Firstly,in the task scheduling stage,the instability of mobile devices leads to low analysis efficiency.Secondly,in the task execution stage,the dynamic execution of device-side code has poor scalability and compatibility,and there is a risk of privacy leakage.In view of this,this paper focuses on Android devices,conducts in-depth research on the above problems,and proposes effective solutions.The specific contributions of this paper are as follows:(1)Research on device scheduling algorithm based on response-data statistical models.Although the online data analysis mechanism for mobile devices can protect data privacy and security,the instability of the mobile environment exacerbates the "long tail”phenomenon of distributed data analysis.To address this issue,this paper proposes a device scheduling algorithm based on response data statistical models.The algorithm establishes a mathematical model based on historical data without requiring prior information and optimizes device scheduling.Experimental results show that the algorithm can effectively achieve joint optimization of task query latency and device resource consumption.(2)Research on the user-level Android execution sandbox mechanism.To address the low scalability and compatibility of existing data analysis paradigms and to avoid privacy breaches caused by unexpected information access,anew online data analysis paradigm based on dynamic code execution and an Android runtime sandbox mechanism for user-level applications are proposed.The mechanism uses system call interception policies to achieve execution isolation for dynamic modules in applications.Experimental results show that this mechanism can effectively avoid unauthorized access during dynamic code execution while maintaining low device resource consumption.(3)Design and implement a privacy-protected online data analysis engine.The engine integrates the above research results and provides data analysts with quick and efficient online data analysis capabilities while protecting user privacy.Finally,this paper installed a test application on 1642 volunteer mobile phones and conducted systematic testing of the online data analysis engine.The test results show that the system can provide low-latency online data analysis capabilities for data analysts while effectively protecting device data privacy and security.