Innovative NIDS with Optimized Deep CNN for Feature Engineering

The Internet of Things is integrated into many automated network systems. The Internet of Things has brought about significant operational shifts in a number of companies during the last several years. Many IoT nodes gather, store, and analyze information that may be used to discover someone's identity. Because of this, cybercriminals find them easy to attack. Researchers have investigated this problem and made several systems to find intrusions. With how things are done now, improving performance and discovering new forms of cyberattacks are hard. In this study, a DCNN-based IDS is recommended. A DCNN has three layers of dense connections and two layers of convolution. Proposed The method that is recommended improves performance while using less computing power. With the IoTID20 dataset, tests were done. The effectiveness of the proposed model was evaluated based on its accuracy, recall, precision and F1-score, among other performance measures. In the proposed model, different optimizing methods were used to ensure that Nadam, Adam, and SGD worked as well as they could. Also, both contemporary deep learning and more conventional machine learning approaches served as benchmarks against which the proposed model was estimated.