Leveraging Metaheuristic Algorithms for Optimal Feature Selection in IoT Cybersecurity: A Study on Enhancing DDoS Attack Detection

In the realm of interconnected devices, such as Internet of Things (IoT) networks, vulnerabilities can often lead to the manifestation of Denial of Service (DoS) and Distributed Denial of Service (DDoS) attacks, thereby posing significant challenges to the robustness and reliability of these systems. These attacks engage the server or multiple components of the network with flooded traffic which is mostly artificially generated. This does not allow for user requests to be responded to by the server, hence reducing the performance of the network. Metaheuristic algorithms have emerged as crucial tools in addressing optimization problems, with many of these approaches drawing inspiration from the collective intelligence and foraging behaviors exhibited by natural organisms, thereby providing novel and effective strategies for identifying optimal solutions. This paper introduces a nature inspired metaheuristic Salp Swarm Algorithm (SSA), inspired by the collaborative foraging and navigation behaviours of Salps. Accuracies are validated by various classifiers like CNN, RNN, DNN. Using metaheuristic algorithms like SSA for optimal feature selection and deep learning classifiers to classify over which protocol the attacks are taking place, we are able to predict which type of attack (DDos or DoS) takes place given the important features shortlisted using the feature selection algorithm. RNN gives better accuracy of 100% on BoT-IoT datasets.