A Novel Hybrid FSO-SVM Model for Attack Detection and Classification in Social IoT

Abstract

Modern Society has multiple channels of communication therefore, multiple ways are there to promote sociability and social relationships which offer the construction of social identities. Among all of these methods, the internet seems to be a potent instrument for modern society's communication. The term "Social Internet of Things" infers to a novel strategy that applies the Social Network Paradigm to the Internet of Things (IoT) domain, facilitating communication while enhancing the relationship between users and devices. A secure communication can be provided by using a novel hybrid classification approach which is developed by combining the benefits of SVM and FSO. FSO is an innovative approach that combines AI strategies with clarified and enhanced security processes. SVM stands out as an essential and popular classification method. SVM performance is highly dependent on choosing the most important characteristics and determining kernel settings effectively. The FSO procedure is also notable for updating positions through element-wise Hadamard matrix multiplication processes. By allowing for simultaneous processing on several data items, this operation shortens the computation time overall. Microsoft Research Paraphrase Corpus datasets are used to evaluate the proposed model, and compared with several well-known metaheuristic algorithms that have used to enhance the performance of SVM. Many attack datasets are used in widespread experiments to verify this system. The results are compared based on performance metrics such as accuracy, precision, recall and F1-score. The proposed framework gains a 99.2% overall classification accuracy.