A Lightweight Convolutional Neural Network based network intrusion detection and classification method for Social Internet of Things

Abstract: The Internet of Things (IoT) and social networking principles have been combined to create the so-called Social Internet of Things paradigm, which holds that devices can build social connections with their owners on their own. In this scenario, "things" engage with their peers in order to find the services they require.If the IoT is not handled properly, assaults and issues could overshadow any advantages.The quantity and methods of attacks have, however, expanded due to the extraordinary improvement of this technology. The complexity of maintaining data privacy as a result makes it even more challenging to offer top-notch services and complete security.The development of intrusion detection systems (IDS), which can quickly and accurately identify and categorise intrusions at various levels of networks, has made substantial use of deep learning techniques. In this work, a network intrusion detection model based on convolutional neural networks that has five convolutional layers is proposed. This model is tested using the CICIDS2018 dataset, a publicly accessible dataset with 80 statistical features, for both binary and multi-class classification. Data transformation and numerical standardisation procedures are used to pre-process the dataset.Experiments are conducted to assess the performance of the proposed system, and the results of the study demonstrate that the proposed CNN outperforms existing intrusion detection techniques in terms of multi-class categorization detection, with average values for accuracy, precision, recall, and F1-score of 99.65%, 99.16%, 98.70% and 99.09%, respectively.