Encryption represents a fundamental tool to protect the confidentiality of sensitive data, such as identity attributes, medical data, and digital twins of physical objects. By employing encryption, sensitive data can be securely stored in not-fully-trusted cloud systems, thereby enabling various use cases.
However, encryption also introduces challenges: If decryption keys are lost (e.g., because the device holding them breaks), users face the threat of losing access to their encrypted data, as they are no longer able to decrypt these data. Furthermore, while encryption prevents the cloud from learning the users’ data, the cloud is also greatly hindered in processing the encrypted data even if the user would benefit from the results.
We investigate the two challenges caused by encryption and propose solutions in the context of a digital twin system. A digital twin is the continuously updated digital representation of a physical object that is maintained in the cloud. As digital twin data can be highly sensitive, we have previously proposed a security architecture for digital twins that introduces an encryption layer to protect the data. For such an encrypted digital twin system, recovery and processing are also core requirements. Firstly, we make use of the flexibility of proxy re-encryption to change access permissions dynamically (e.g., as trust relationships change) and recover digital twin data on a replacement device after the original device is no longer functional. Furthermore, we integrate secure multi-party computation to process protected data according to the users’ permissions without revealing the inputs or outputs to the processing nodes. An evaluation highlights the feasibility and practicability of our approach based on an example use case, namely privacy-preserving contact tracing.