Improving the Accuracy of Network Intrusion Detection System in Medical IoT Systems through Butterfly Optimization Algorithm

Network intrusion detection systems analyze traffic in a medical IoT system to detect abnormal behaviors. Machine learning and artificial intelligence (AI) algorithms are widely used in designing intrusion detection systems to prevent attacks on a medical IoT system. In this paper, an artificial neural network is employed to detect abnormal behavior in a medical IoT system. The accuracy of the detection depends heavily on the features that are fed into the artificial neural network. Selecting the important and discriminative features of network traffic is a crucial and challenging issue because it has a significant impact on the learning process. In the proposed method, the butterfly optimization algorithm which is a meta-heuristic optimization algorithm is employed to select the optimal features for the learning process in an artificial neural network. The results achieved, 93.27% accuracy, indicate the capability of the butterfly optimization algorithm to determine discriminative features of network traffic data. The proposed algorithm outperformed the decision tree, support vector machine, and ant colony optimization, which was proposed in previous researches for the same goal.