With the advancement of IoT technology, an increasing number of devices and sensors are connecting to the network, leading to data fragmentation due to diverse data standards adopted by different enterprises. Typically, IoT data is stored on company servers and is at risk of being lost if non-profit projects are terminated. Moreover, due to their size and prominence, large enterprise servers often become primary targets for hackers, posing threats to user privacy and security. In this study, we designed and implemented the decentralized IoT data platform, UPOD. Adhering to a human-centric approach, UPOD aims to provide users with a platform for personal online data storage, empowering them to self-manage their data. We advocate for the use of the ActivityPub protocol, commonly applied in decentralized social networks, as UPOD's communication protocol to unify the management of UPOD and social network data. Recognizing the diverse nature of user data, we introduced the concept of virtual roles and developed an IoT data allocation mechanism. This allows users to customize data assignment to various virtual roles and decide on sharing this data with third parties. To demonstrate its feasibility, we constructed a UPOD prototype for evaluation. The UPOD system offers guidance in addressing the issue of IoT data fragmentation. The introduced concept of virtual roles allows precise control over which data is shared or revoked, ensuring only appropriate or necessary data is shared or withdrawn, safeguarding user privacy. Since UPOD provides unified storage for users' personal data, their information remains accessible even if a company's service or project ceases, ensuring the security of individual IoT data.