Triple methods-based empirical assessment of the effectiveness of adaptive cyber defenses in the cloud.

The flexible and cost-effective service provided by cloud computing has led to the development of a vast array of applications in smart cities. Nevertheless, their traditional security approaches presented the concept of a static target for attacks, leading to an asymmetric situation between defenders and attackers. Adaptive cyber defense (ACD) has, therefore, recently emerged as a game-changer to reverse this asymmetry by reconfiguring the system according to the network scenario. Analyzing and quantifying the effectiveness of these ACDs are of high importance. Previous research on ACD analysis focused more on either studying the system properties using experiment-based approaches or on evaluating its effectiveness by different mathematical modeling approaches. However, little effort has been made to overcome the problems of isolated solutions. In this paper, we described the defensive process as a racing game between the attacker and the defender. On this basis, we conducted a thorough ACD effectiveness evaluation and suggested a comparison strategy using three methods, namely semi-Markov, stochastic reward net, and experimental methods. The obtained simulation results were approximate, and the error rate was less than 3.36% reflecting the reliability of the proposed methods. Based on the assessments, we finally summarized the features of these methods to specify and deduce different scenarios and their corresponding suitable evaluation method. [ABSTRACT FROM AUTHOR]