Supporting Secure Provenance Update by Keeping “Provenance” of the Provenance

Part 2: Asian Conference on Availability, Reliability and Security (AsiaARES)International audienceProvenance of data is a documentation of the origin and processes that produce the data. Many researchers argue that the provenance should be immutable: once a provenance is submitted, it should not be changed or updated. A main reason is that the provenance represents the history of data, and the history should not be altered or changed because it represents the fact in the past. Provenance can be represented by a graph, where each node represents the process executed by a party and an edge represents the relationship between two nodes (i.e. a child node uses the outputs of the parent nodes). A method to ensure that the provenance has not been updated is by using signature chain, where the signatures of the parent nodes are recorded in the children nodes so that any changes to the parent nodes will raise inconsistencies between the parent and the children. However, sticking to the requirement that the provenance should be immutable requires unlimited data storage and also we have problems whenever we need to update the provenance for an accidental error. In this paper, we propose a method that allows updates in the signature chain-based secure provenance, while keeping the signature consistent. The main idea is by keeping the ”provenance” of the provenance itself, that is the history of update of the provenance, in the form of the signatures of the previous versions of the nodes. We implement the idea by keeping the signatures of the previous version in a signature tree similar to the Merkle-tree, where the a parent node in tree is the aggregate signature of the children. Using this method, the storage requirement to store signatures is always smaller than the number of updates