Using Diversity to Evolve More Secure and Efficient Virtual Local Area Networks
Ariel José Aizpurúa, Errin W. Fulp, Daniel A. Cañas · 2023
Virtual Local Area Networks (VLANs) are often established based on a security policy enumerating permissible communications. Unfortunately, finding an efficient (small) number of VLANs to provide only the permissible interconnections becomes more problematic as the number of devices and the complexity of the security policy increases. Given the difficulty of the problem, Genetic Algorithms (GAs) have been used to search for appropriate VLAN configurations; however, the approach can converge to configurations that are either inefficient and/or insecure.This paper describes a new GA-based approach for discovering VLANs that avoids early convergence by monitoring the diversity of the current solution set (GA population). When necessary, the approach adjusts solutions to consider a larger number of VLANs while maintaining the same level of security. Simulation results indicate this approach performs better than other contemporary GA-based techniques. It is able to consistently find concise and secure VLAN groups under various conditions, including an increasing number of device interconnections.