COTRA: A Data Trading Framework for Multi-Source Data Cooperation

Data trading significantly enhances data-driven technologies by enabling efficient data sharing across mobile devices and communication systems. Despite the clear advantages of incorporating multiple sources often experienced by data users, the topic of multi-source data trading remains largely underexplored. This paper designs a data trading framework that enables multi-source data trading through structured and adaptive cooperation among data sources. The proposed framework aims to enhance data usage efficiency and seller revenue. Key technical challenges addressed include the complex interactions between data sources and buyers, the coupling among diverse data products, and the inherent non-convexity in optimization problems. We employ a Nash bargaining framework to model cooperative seller decisions and a two-stage Stackelberg game for dynamic seller-buyer interactions. By systematically addressing the coupling among data products, we derive closed-form solutions despite the non-convex nature of the optimization problem. Our results reveal that seller revenue remains stable with increasing product coupling until it reaches a threshold, beyond which further coupling increases revenue due to the substitute effect. Adaptive cooperation exhibits greater scalability and resilience, particularly in environments with numerous data sources and complex product interdependencies, while structured cooperation provides more predictability and efficiency in stable conditions. Experimental results demonstrate that this framework can improve seller profits by up to 46.32% compared to existing data trading methods in the current market.