AResNet Model Using Deep Learning Approach for Enhancing the Internet of Things (IoT) Forensic Readiness Framework

The rapid growth in the usage of the Internet of Things (IoT) has expanded both in size and scope.The automation and real-time services provided by IoT devices have increased their vulnerability to ongoing attacks.One of the significant challenges in forensic analysis is the limited computational capability and storage capacity of IoT devices, which complicates the process of collecting and identifying various attack types during forensic readiness investigations process.This research proposed an Advanced Residual Network (AResNet) as an integrated model for all IoT devices, incorporating performance improvements from previous research to enhance the IoT forensic readiness framework.Explains the phase of collecting attack data into a forensic repository to identify and analyze types of attacks on IoT networks using a deep learning approach.The proposed forensic IoT readiness framework has three novel functions.First, the repository serves as a secure storage solution for data on various attacks targeting IoT devices, ensuring data integrity and providing a reliable reference source for valid digital evidence.Second, utilizes Particle Swarm Optimization (PSO) algorithms to determine the best and optimal hyperparameters, resulting in a significant improvement in the performance of the deep learning model.Third, it involves the development of the Advanced Residual Network algorithm based on deep neural networks with the collective intelligence of PSO, functioning as an Early Warning System.This system is designed to identify various types of attacks on different IoT devices with a high degree of accuracy.The AResNet model was trained and evaluated using the N-BaIoT dataset, and its performance was compared with various other deep learning models, including RNN, LSTM, CNN, 1D-CNN, and CNN-LSTM.The experimental results clearly demonstrate that the proposed model significantly outperforms previous methodologies in identifying and classifying types of attacks across all IoT devices directly, achieving an accuracy of 0.92, precision of 1.00, and recall of 1.00.Consequently, having a greatly impact on the effectiveness of the IoT forensic readiness investigation process.