An efficient cyber‐physical system using hybridized enhanced support‐vector machine with Ada‐Boost classification algorithm

The necessity of cyber-security has obtained immense importance in day-to-day concerns of network communication. Therefore, several available research works predominantly focus on network security to protect the resources, services, and networks from any unauthorized access. A CPS (cyber-physical system) model using a dual mutation-based genetic algorithm, with feature classification through Ada-Boost and SVM classifier is proposed in this paper. Dual-mutation based genetic-algorithm overcomes the issues of conventional techniques including convergence issues and local fine-tuning of features. In this paper, necessary modifications were made to the existing Genetic Algorithm (GA) method to reduce the random nature of the traditional GA method. Particularly, the goal of this work is to develop the modified reproduction operators with appropriate fitness functions to guide simulations to gain optimal solutions. In floating-point representation, every chromosome vector has been coded as a floating-point number vector having the same length as the solution vector. Each element was selected initially, to stand within the desired domain, and operators were designed carefully in satisfying the constraints. As a result, there are various enhancements employed in the dual-mutation algorithm that handles local fine-tuned features. The relevant features of dataset samples are extracted and rescaled using feature selection and resampling phase aided by the Markov-resampling process. Followed by this, a hybrid approach of ESVM (enhanced support-vector machine) algorithm with Ada-Boost classifier is implemented for the fault classification process. The performance assessment was explicated in terms of accuracy-factor, F1-score, and execution time. Comparative analysis expounded the efficacy of the proposed model than other conventional methods attaining higher accuracy (97%), F1-score (99%) rates, and less execution time (15.33 s).