Design of an Efficient Forensic Layer for IoT Network Traffic Analysis Engine Using Deep Packet Inspection via Recurrent Neural Networks

Abstract

With the rapid proliferation of Internet of Things (IoT) devices, the security and integrity of network traffic have emerged as critical challenges. The exponential growth of IoT devices has introduced complex security vulnerabilities that demand innovative solutions. Analyzing IoT network traffic and detecting attacks in real-time present formidable challenges. Traditional security measures often fall short in addressing the adaptable and dynamic nature of these threats. The below paper presents a new Deep Packet Inspection technique using a combination of Recurrent Neural Networks, LSTM, and GRU. Using DPI, the facility can be made available to extract and analyze parameters like protocol, source, destination addresses, port numbers, payload, timestamp, packet length, sequence number, flags, quality of service markings, content type, content length, user agent, referrer metric parameter sets. The accuracy and intensity of the detection results for the attacks imposed in the network traffic data are enhanced with LSTM and GRU architectures. Formidable robustness in detecting the imposed attacks was determined to improve security in the IoT forensic layer while analyzing the network traffic. Usability can be applied in real-time monitoring systems, intrusion detection and prevention systems, and forensic investigation. For example, it ensures protection for sensitive data. It would allow connected devices and services to run without disturbance through the targeted detection of specific attacks like DoS attacks, malware exploitation, and unauthorized access attempts. To conclude, the outline of this paper falls within the scope of some of the matters that must be dealt with promptly related to the security of IoT networks through a remarkable innovative solution, that is, the usage of DPI and RNNs based- LSTM and GRU network architectures. The obtained results related to the following factors show not just good precision and good accuracy but also good recall, which showed high confidence in detection.