This thesis aims to advance the field of physical layer security across multiple communication systems, including power-limited devices, Orthogonal Frequency Division Multiplexing (OFDM), precoded spatial modulation systems, Unmanned Aerial Vehicles (UAVs), and Intelligent Reflecting Surface (IRS) assisted Non- Orthogonal Multiple Access (NOMA) systems. The research encompasses the development and analysis of various innovative approaches that exploit inherent characteristics of these systems to improve their security. The first focus is the establishment of secure data transmission protocols for power-limited devices. A secure turbo code design is proposed that effectively combines encryption and error correction, providing enhanced security without increasing complexity. The potential of OFDM as a source of randomness for physical layer security designs is explored next. A pre-shared key-based algorithm is introduced to alter the constellation mapping of modulated symbols, making cryptographic attacks more challenging while maintaining transmission quality. Subsequently, a combined precoding and Artificial Noise (AN) insertion scheme is proposed to elevate the secrecy of precoded spatial modulation systems, even when the wiretap channel is partially correlated with the legitimate channel. This scheme capitalizes on the legitimate Channel State Information (CSI) without requiring knowledge of the eavesdropperâs channel. The security of UAVs receives particular attention as the physical layer attributes are exploited to deter control and spoofing attacks. Then a deep learning-based localization verification system is developed to authenticate received signals with high accuracy. Finally, the thesis presents a secure precoding technique for IRS-NOMA systems. The proposed method, guided by a Deep Deterministic Policy Gradient (DDPG) reinforcement learning algorithm, harnesses phase shift and user-channel characteristics to safeguard user privacy while preserving interference cancellation capabilities.
|Date of Award
|31 Dec 2023
- The University of Manchester
|Laith Danoon (Supervisor) & Emad Alsusa (Supervisor)