Machine learning based privacy-preserving fair data trading in big data market

Abstract

In the era of big data, the produced and collected data explode due to the emerging technologies and applications that pervade everywhere in our daily lives, including internet of things applications such as smart home, smart city, smart grid, e-commerce applications and social network. Big data market can carry out efficient data trading, which provides a way to share data and further enhances the utility of data. However, to realize effective data trading in big data market, several challenges need to be resolved. The first one is to verify the data availability for a data consumer. The second is privacy of a data provider who is unwilling to reveal his real identity to the data consumer. The third is the payment fairness between a data provider and a data consumer with atomic exchange. In this paper, we address these challenges by proposing a new blockchain-based fair data trading protocol in big data market. The proposed protocol integrates ring signature, double-authentication-preventing signature and similarity learning to guarantee the availability of trading data, privacy of data providers and fairness between data providers and data consumers. We show the proposed protocol achieves the desirable security properties that a secure data trading protocol should have. The implementation results with Solidity smart contract demonstrate the validity of the proposed blockchain-based fair data trading protocol.