Preserving integrity and confidentiality of a directed acyclic graph model of provenance

Abstract

This paper describes how to preserve integrity and confidentiality of a directed acyclic graph (DAG) model of provenance database. We show a method to preserve integrity by using digital signature where both of the provenance owner and the process executors (i.e. contributors) sign the nodes and the relationships between nodes in the provenance graph so that attacks to integrity can be detected by checking the signatures. To preserve confidentiality of the nodes and edges in the provenance graph we propose an access control model based on paths on the provenance graph because an auditor who need to audit a result normally need to access all nodes that have causal relationship with the result (i.e. all nodes that have a path to the result). We also complement the path-based access control with a compartment-based access control where each node is classified into compartments and the auditor is not allowed to access the nodes included in a compartment that can not be accessed by him/her (because of the sensitivity of the compartment). We implement the path-based access control by encrypting the nodes and later store encrypted encryption's keys in the children of the nodes. The compartment-based access control is implemented by encrypting the nodes in different compartments with different keys. We developed a prototype of the model and performed experiments to measure the overhead of digital signature and the double encryptions. © 2010 IFIP International Federation for Information Processing.