An access control model considering with transitions of access rights based on the blockchain
Hirotsugu Kinoshita, Morizumi Tetsuya · 2022 IEEE 46th Annual Computers, Software, and Applications Conference (COMPSAC) · 2022
Leakage of sensitive information concerned with database systems, such as social networking services and online storages, has occurred. One of the causes of such leakage is a covert channel, which is a path of information leakage caused by inconsistencies in access rights on databases. Another cause is an inference attack in which secret information is inferred from the fragments of open data. We previously proposed a hypergraph-based model used to prevent the possibility of information leaks by inference but considered only reading operations. However, writing operations also have to be considered. We also previously proposed covert-channel analysis models with inference rules and reading and writing operations. Management mechanisms for access histories to objects by users are required to analysis the satisfaction of the inference rules. If the accesses histories are altered, information leakage by an inference attack will occur. In this study, we used a blockchain to manage access logs to solve the above problems. A traditional covert channel is a static analysis based on the assumption that the access rights of access control lists do not change. Covert channels exist under certain conditions even if no covert channel is detected from static analysis. We define the authorization quad, which describes the relation among the subject as a user, object as a file, access rights, and timing of transition, to consider the transitions of access rights to analyze a covert channel. We also propose a management scheme for covert-channel analysis with a blockchain considering inference attacks and transitions of access rights.