A Game Theoretic Approach for Data Asset Protection in Metaverse

The metaverse maintains a virtual space in which users can interact through their avatars. To facilitate the virtual activities, users could bring their real-world data to the virtual world, which constitutes their data assets; users (or avatars) therefore can purchase data from others to feed their metaverse applications, as well as sell their data assets to make profits in the metaverse. The sharing of data assets would proliferate the metaverse applications (e.g., AI training) and enrich the user experience, which however would also disclose the user privacy if inappropriately used. In this paper, we study the data asset protection in the metaverse to prohibit dishonest data asset distributions. Specifically, to make a profit, a user may purchase data assets and then resell the data to others; the resale behaviours violate the agreement of data asset trade and harm the original data owner. To prevent this behaviour, in this work, a user is dictated to deposit a certain amount of credits to a third authority when purchasing data. The deposit would be released and returned to the purchaser if no dishonest behaviours are observed; otherwise, the deposit would be paid to the data owner to compensate for its loss. This deposit amount should be carefully balanced as overwhelming deposits would increase the burden of data asset purchase, whereas a trivial amount would not be able to prohibit dishonest behaviours. We model the problem from a game theoretic perspective and develop a distributed solution to strike a balance. Using analysis and simulations, we prove the existence of the Nash Equilibrium, and show that our scheme can effectively reduce the risk of dishonest data asset distribution.