Skip to Main Content
This paper presents an access control algorithm which bases on artificial neural network (ANN). It uses selected roles as input vectors. Then, considering role inheritance, the matching roles that may mutual exclude are picked as the output vectors to train the role-role ANN. After that, it removes all conflicts from the output of the former ANN and makes them as the new input vectors. And it employs users' final permissions as the output vectors to train the role- permission ANN according to users' current sessions. This algorithm has high efficiency and can assign users various permissions in different session. It need not re-compute trained ANN if we add or delete users in the system without changing the mapping between the roles and permissions. It exploits bit strings to express roles and permissions, which reduces the data transmission and fits for low bandwidth networks.