Document Details

Document Type : Thesis 
Document Title :
DEVICE-TO-DEVICE CONTINUOUS AUTHENTICATION USING MACHINE LEARNING FOR THE INTERNET OF THINGS
عنوان الرسالة: المصادقة المستمرة بين الأجهزة باستخدام تعلم الآلة لإنترنت الأشياء
 
Subject : Faculty of Computing and Information Technology 
Document Language : Arabic 
Abstract : Device-to-device (D2D) authentication is a fundamental security requirement that cannot be neglected in Internet of Things (IoT) environment. Usually, devices are authenticated statically at the beginning of the session. Session impersonation is one of the issues that arise with static authentication. Continuous authentication, in which the entity is continuously authenticated at a given frequency throughout a session, is an efficient solution that overcomes this vulnerability. Artificial intelligence in IoT security solutions safeguards communications and improves the identification and prediction of attacks. Coupled with edge computing, it can result in improved performance and decreased latency. However, edge-based D2D continuous authentication with machine learning is still in its early stage. In this thesis, we provide a comprehensive review of the IoT environment, authentication in IoT, and authentication challenges. Furthermore, we investigate wireless device signatures that are robust against rogue agents, namely Radio Frequency Fingerprinting (RFF) to identify devices. We propose a D2D continuous authentication model for IoT devices that uses deep learning and RFF to detect illegitimate devices. The transmitted radio frequency signals are fed frequently during the session to the deep learning model to check the transmitter device's legitimacy. Experimental results demonstrate that our model performance, in terms of accuracy, precision, recall, and F-score, achieve 99.65%, 1.0, 99.30%, and 99.64% respectively. The process of detecting a frame group takes into account environmental conditions, real-time sensing, and validation. This reduces latency so that the process of verifying each frame group takes less than 0.019s. 
Supervisor : Dr. Suhair Alshehri 
Thesis Type : Master Thesis 
Publishing Year : 1444 AH
2023 AD
 
Added Date : Wednesday, August 9, 2023 

Researchers

Researcher Name (Arabic)Researcher Name (English)Researcher TypeDr GradeEmail
أسماء مسعد السفريAlsefri, Asmaa MasadResearcherMaster 

Files

File NameTypeDescription
 49277.pdf pdf 

Back To Researches Page