Efficient Feature Selection for IoMT Security: Using Artificial Bee Colony for Intrusion Detection
Divya K, V. Nagaraju · 2024
Internet of Medical Things (IoMT) is refacing healthcare through patient monitoring, remote diagnosis and better informatics in clinical decision making. However, this dependence increases risks as IoMT networks become vulnerable targets of cyber-crimes that seek to compromise the patient data, and the functionality of devices. Therefore, there is need to have strong Intrusion Detection Systems (IDS) which should protect these networks. Although several IDS models have been proposed in the literature, a limitation of most IDS models when applied to high dimensional IoMT data is that they do not perform well in real-time because they are plagued by high false positive rates and long detection times. Towards this end, this paper presents a ABC-DBN model for feature selection employing the Artificial Bee Colony algorithm and intrusion detection using a Deep Belief Network. A feature selection approach of the ABC algorithm, helps in reducing the dimensionality of the data set, thus minimizing noise and computational costs. The ABC-DBN outcomes reveal that the proposed model yields nearly optimal classification solution with the AUC 1.0, high precision and shorter convergence time and give confirmation of superior performance in comparison with conventional IDS algorithms. These results confirm that the model is well placed for real-time scenarios as it can provide good intrusion detection while requiring little computational effort. Hence, the proposed ABC-DBN model present a feasible solution to addressing the secured IoMT and the problem of handling high dimensions data and real-time detection. This approach holds much promise in enhancing the protection of data in IoMT settings and outperforms existing methods substantially.