| Summary: | <p>Modern methods to solve the practical problems related to the information security system design in multipurpose distributed systems often use optimization models. The paper formulates a problem statement of bilinear Boolean programming with the quality score that determines the level of the system security to select classes of protected objects in distributed systems and allocate databases in these objects. Restrictions specify the maximum cost of protection and the recommended amount of data stored in the object. To solve this problem, using the algorithms of discrete programming is possible, but the accurate algorithms, generally, are exponentially timeconsuming. To find a solution the paper proposes to interpret this problem as a problem of two players with non-conflicting interests. The first player is responsible for the assignment of security classes for the system objects, and the second player is responsible for the distribution of databases between the objects. To find solutions the use of a Nash equilibrium criterion is offered.</p><p>To find the equilibrium solution is developed an algorithm and are proved its convergence, and the fact that it really allows us to obtain the Nash equilibrium position. Using the problem-solving approach related to reducing the original optimization problem to a two-player game with nonconflicting interests allowed a decreasing dimension of the original problem, because at each step of the algorithm is solved the problem of discrete optimization of a smaller dimension than the original problem of Boolean programming.</p><p>The paper presents results of solving the problem, as well as research results on the time complexity of represented algorithms. It conducts a comparative analysis of the two approaches: an approach based on solving a discrete optimization problem and an approach based on reducing the problem to a two-player game with non-conflicting interests. An approach based on reducing the initial problem to a game, allowed us to decrease solution time significantly, by 2-3 orders, while, in general, the reached value was the same as in the original problem of Boolean programming.</p>
|
|---|
