Administrating role-based access control by genetic algorithms

Abstract

In the paper we address the problem of administering Role-Based Access Control (RBAC) systems, which consists in substantiation of the choice of solution of the RBAC design or reconfiguration problem based on using genetic algorithms (in conditions of dynamically changing access control policies). The problems of RBAC design and reconfiguration are NP-complete. Therefore, the use of genetic algorithms to solve it seems to be quite appropriate. The paper discusses the mathematical basis for selection of administrative tasks and novelties, implementation of which allows to increase the speed of the developed genetic algorithms. We also consider the structure and possibilities of the developed testbed and the results of its application to estimate the administrative costs to implement changes to an access control policy using the solutions of administration tasks. Additionally, using the testbed we evaluated the gain in speed of the genetic algorithms that was obtained by implementing the proposed contributions, such as creating two chromosomes for each individual, using columns of Boolean matrices as genes of chromosomes, as well as additional procedures that improve the operations of crossover, mutation and selection. The results of the experiments showed high efficiency of the proposed algorithms.