| In October 2019,the Fourth Plenary Session of the 19th Central Committee of the Communist Party of China added "data" as a factor of production for the first time,and then pointed out that it is necessary to accelerate the construction of a data market around data elements to help the country occupy the commanding heights of global competition in the digital economy.At the same time,with the development of mobile networks,social networking applications,and intelligent world systems,a large amount of data is being collected.However,this data is stored in the databases of organizations or companies,creating a series of "data silos".The application of new technologies,such as machine learning and data mining,requires the integration of multiple data sources to unlock the value contained in the data.Therefore,the circulation and sharing of data is the core content of data market cultivation.However,public welfare or free data sharing is difficult to achieve in reality,and economic-driven data transactions are becoming an effective way to achieve data sharing.Through data marketplaces,data owners can sell their data for fair remuneration,and data consumers can buy massive amounts of data that meet their business needs,thereby improving the quality of their research,products,or services.Therefore,data transactions are a win-win process for data owners and consumers.However,as a virtual commodity,data,with its variable,diverse,massive,high-speed and complex basic characteristics,has brought challenges to the current trading model that mainly faces physical commodities.Therefore,to realize the true value and utility of big data,the traditional transaction model must be re-evaluated and improved.At the same time,data also has the characteristics of easy leakage,low replication cost,and rich privacy information,which has more urgent needs for transaction security and privacy protection.Therefore,how to build an effective data trading model and ensure the security and privacy protection of the trading is a problem that needs to be studied.Under the existing foundation and conditions,this paper focuses on the optimization of data trading and the privacy protection issues in the trading process.Through in-depth research on the data trading model from centralized single-objective optimization to distributed single-objective optimization,and then to distributed multi-objective optimization,to solve the above-mentioned trading optimization and privacy protection problems.The main research contents and innovative work of this paper are as follows:(1)Research on secure data trading model for platform profit optimizationAiming at the problem that the current data trading platform may leak the data privacy of the data owner,a secure data trading model for platform profit optimization is proposed.After the data owner collects the data required by the data consumers,he will analyze the data locally,and then send the analysis results to the platform after adding noise according to the privacy budget allocated by the platform to protect his data privacy.At the same time,in order to encourage data owners and data consumers to actively participate in data trading,a trading optimization model is constructed to optimize the platform’s profit,taking into account the consumer’s tolerance constraint on the result deviation and the owner’s privacy leakage compensation.The model is solved by a profit optimization algorithm based on genetic algorithm.In the profit optimization algorithm,the data trading platform generates populations according to the tolerance constraints and compensation information submitted by consumers and the quotations of data owners.In order to maximize its profit,the platform calculates the optimal result through the selection,inheritance and variation of the populations.The simulation results show that the algorithm not only protects the data privacy of the data owner,but also the profit of the data trading platform can be further improved under this method.(2)Research on secure data trading model for system utility optimizationAiming at the risk of data privacy leakage in the centralized data transaction scheme,and the current research only considers the benefits of the data trading platform,and has limitations in the overall utility of the system,a secure data transaction model for system utility optimization is proposed.This model realizes peer-to-peer data transactions between data owners and data consumers,and solves the crisis of data privacy leakage in centralized solutions.At the same time,a data transaction optimization model with the goal of maximizing system utility is constructed.In order to solve the model while protecting the privacy of sensitive personal information of data consumers and data owners during the solving process,a double auction mechanism is proposed.In this mechanism,the data agent calculates the data trading amounts according to the quotation information submitted by the data consumers and the owners,and then,the consumers and the owners update their quotations according to the data trading amounts.After multiple rounds of iterations,the transaction plan is determined.The mechanism can solve the model only through the quotations provided by the data consumers and owners,thus avoiding the disclosure of their sensitive personal information.In order to encourage data owners to provide high-quality data and efficient data transmission services,a reputation model is built,and a Proof of Reputation(PoR)consensus mechanism is proposed.Theoretical analysis and simulation results show that the mechanism not only protects the privacy of sensitive personal information of data owners and consumers,but also has better performance in terms of social welfare.(3)Research on secure data trading model for multi-objective optimizationAiming at the security crisis faced by the centralized data trading scheme and the lack of efficiency of the current scheme,and the current research only considers the data itself and ignores the consumer demand factor,a multiobjective optimization-oriented secure data trading model is proposed.The use of consortium blockchain realizes peer-to-peer data trading,which not only ensures the data privacy of data owners,but also improves the throughput and efficiency of transactions.By comprehensively considering the factors of the data-related and consumers’ demands,a bi-layer multi-objective optimization model is constructed based on multi-dimensional factors such as data quality,data attributes,attribute correlation,and consumer competition,which further optimizes the utilities of data owners and data consumers.In order to solve this model and protect the privacy of sensitive personal information of data owners and consumers during the solving process,an improved multi-objective genetic algorithm-cooperative NSGAII is proposed.The algorithm solves the model through the cooperation of data owners,data consumers and data agents.The data owners and data consumers complete the calculation of their respective utility functions locally,and the data agent completes the non-dominated sorting of the population,thus avoiding the leakage of sensitive information.Theoretical analysis and simulation results show that the proposed algorithm not only protects the personal sensitive information of data owners and consumers,but also achieves better performance in terms of their utility. |
PDF Full Text Request