A Network Based Intrusion Detection System For Internet Of Medical Things Using Deep Learning Approach

Due to the technology’s rapid development overall, humans are expected to live alongside it very soon. Because of the development of information and communications technology, all of these technologies are within the means of the average person (ICT). New computer and communication environments, including the Internet of Things (IoT), appeared as a result of this development. IoT research is actively being conducted, which benefits society’s general development and improves people’s quality of life. When it comes to network security, network intrusion detection systems (NIDSs) are superior to more established network defensive solutions like firewalls. The effectiveness of the algorithms and techniques employed to boost classification accuracy and reduce training and testing times is a key factor in NIDS’s success. In order to create a versatile and effective intrusion detection system (IDS) to recognise and categorize unexpected and unforeseen cyberattacks, a convolution neural network (CNN), a deep learning model, is investigated in this research. The suggested method is applied to dimensionality reduction and feature learning. It significantly cuts down on training and testing time while also increasing the effectiveness of support vector machines’ (SVM) ability to forecast assaults. The new characteristics are added to the SVM algorithm after the pre-training phase in order to increase the algorithm’s classification precision and intrusion detection ability. Results show that our method scored better on performance measures for binary and multiclass classification than the majority of the preceding methods. Timeframes for SVM training and testing were also accelerated.