Analysis of Evolutionary Algorithm based Optimization for Cyber Threat Modeling

This study utilizes a type of genetic algorithm (GA) called trait-based heterogeneous populations plus (TbHP+) for evaluating cyber threat model equations. TbHP+ uses immunity and instinct to create a type of memory concept that then offers populations with more efficient directives. Additionally, TbHP+ uses the varying number of individuals based on "traits" and can change them during the search process therefore allowing the size of a population to automatically adapt. The population size is based on specific characteristics such as character fitness and credit for immunity. The TbHP+ model in this study was first tested and compared to the classical GA model regarding minimum error performance for well-known mathematical test functions: Rastrigin's 6th test function, De Jong's 1st Test function, and Schwefel's 7th test function. Further, TbHP+ and classical GA were applied to a cyber threat model, namely the internet worm model. In both analysis, the TbHP+ algorithm performs better than classical GA algorithm, when compared to the minimum error performance in test functions and to the objective function of the internet worm model.