Homomorphic multi-party computation for Internet of Medical Things

Internet of Medical Things (IoMT) has great potential in delivering medical services. In IoMT, data users (e.g., doctors) may want to process data collected by sensors attached to data owners' body (e.g., patients). As sensors lack computing resources, confidential outsourcing the data to a server becomes necessary due to its sensitivity. Using homomorphic encryption raises limitations in secure processing. First, as decrypting the processed result requires the data owners' secret key, they must be online or share it with data users. Second, when processing is performed on the data of multiple data owners, the interaction becomes harder. Finally, if the processed result is sensitive, it lacks confidentiality as data owners may access it. In this paper, we propose a non-interactive homomorphic multi-party computation (HMPC) protocol, addressing the limitations efficiently. In HMPC, data owners encrypt their data with their own key and store it in a cloud server. Then, data users select the required data from the cloud server and outsource their own encrypted data to the server for processing. Afterwards, they decrypt the result regardless of the circuit computed and without interaction with the data owners. Our security and performance analyses demonstrate that HMPC is provably secure and applicable.