A Novel Hybrid Spotted Hyena-Swarm Optimization (HS-FFO) Framework for Effective Feature Selection in IOT Based Cloud Security Data

Abstract

Internet of Things (IoT) has gained its major insight in terms of its deployment and applications. Since IoT exhibits more heterogeneous characteristics in transmitting the real time application data, these data are vulnerable to many security threats. To safeguard the data, machine and deep learning based security systems has been proposed. But this system suffers the computational burden that impedes threat detection capability. Hence the feature selection plays an important role in designing the complexity aware IoT systems to defend the security attacks in the system. This paper propose the novel ensemble of spotted hyena with firefly algorithm to choose the best features and minimise the redundant data features that can boost the detection system's computational effectiveness. Firstly, an effective firefly optimized feature correlation method is developed. Then, in order to enhance the exploration and search path, operators of fireflies are combined with Spotted Hyena to assist the swarms in leaving the regionally best solutions. The experimentation has been carried out using the different IoT cloud security datasets such as NSL-KDD-99 , UNSW and CIDCC -001 datasets and contrasted with ten cutting-edge feature extraction techniques, like PSO (particle swarm optimization), BAT, Firefly, ACO(Ant Colony Optimization), Improved PSO, CAT, RAT, Spotted Hyena, SHO and BOC(Bee-Colony Optimization) algorithms. Results demonstrates the proposed hybrid model has achieved the better feature selection mechanism with less convergence time and aids better for intelligent threat detection system with the high performance of detection.