BWKA: A Blockchain-Based Wide-Area Knowledge Acquisition Ecosystem

Yang Xu,Jianbo Shao,Jia Liu,Yulong Shen,Tarik Taleb,Norio Shiratori
DOI: https://doi.org/10.1109/tdsc.2024.3382031
2024-01-01
IEEE Transactions on Dependable and Secure Computing
Abstract:Benefiting from the booming of Big Data and artificial intelligence (AI) technologies, data-as-a-service is gradually transforming into knowledge-as-a-service. Extracting knowledge from massive raw data is becoming a popular paradigm to save network resources and improve efficiency, and establishing knowledge markets is receiving increasing attention from academia and industry. In this paper, we propose a one-stop knowledge acquisition ecosystem termed BWKA that covers the whole process from upper-layer knowledge trading to underlying knowledge generation. In the knowledge trading process, the knowledge-as-a-service platform (KSP) is the buyer and publishes knowledge demands to multiple local knowledge sellers (LKSs). In the knowledge generation process, each LKS aggregates data from its sensors and then trains data into knowledge according to the KSP's requirements. We resort to blockchain technology and provide a series of tailored operating rules and functions to protect the truthfulness of data gathering and the fairness of knowledge trading. In addition, we introduce incentive mechanisms to stimulate selfish and rational entities in the BWKA ecosystem to participate in knowledge acquisition. To analyze the strategic interactions among entities theoretically, we develop a nested hierarchical game model, where the upper-layer knowledge trading is evaluated based on the Contract Theory, and the lower-layer knowledge generation is formulated as a two-stage Stackelberg game. By solving the nested hierarchical game in a backward inductive way, we identify the optimal strategy for each entity in closed form. Experiments on the Ethereum blockchain and simulation results demonstrate the practical operability and outstanding performance of the BWKA ecosystem.
computer science, information systems, software engineering, hardware & architecture
What problem does this paper attempt to address?