BoTNetShield: Intelligent IoT Security Framework using Convolution Neural Network

The "Internet of Things (IoT)" sector has grown in recent years, penetrating many different markets. The diversity and diminutive size of Internet-connected gadgets render them vulnerable to attacks. Improving and managing network forensic processes to avoid the various attacks, key logging, and denial of service becomes challenging in big and heterogeneous networks. Implementing high-end safety systems is challenging in IoT networks due to their variety and resource restrictions. Using deep learning, botnet detection is made better. It often requires a great deal of internet engagement. Internet of Things devices with little RAM have difficulties when using deep learning. This study use Convolutional Neural Networks (CNNs) for model training and Similarity Associated Feature Selection (SAFS) to decrease the dimensionality of smaller Internet of Things (IoT) traffic networks in order to effectively classify attacks in devices with limited memory. The effectiveness of botnet attacks is evaluated by first collecting datasets from the internet. By improving attack detection data from acquired datasets, data cleaning aids in eliminating unnecessary information. To enhance attack detection, SAFS cleans the data and then extracts efficient deep features. To classify risks to Internet of Things devices based on retrieved deep data, an efficient Recurrent Neural Network method is employed. Research evaluations compare proposed methods against existing ones in order to determine whether one is more effective.