Resource Allocation and Optimization for Secure Wireless Communication Networks

Abstract

This thesis explores challenges in future wireless networks, focusing primarily on resource allocation and optimization within the context of future technologies. The research addresses network security by proposing innovative solutions, especially the utilization of different technologies for physical layer security (PLS) in wireless communication networks across diverse scenarios.
The initial work proposes the integration of blockchain for PLS in cooperative vehicular cognitive radio (CR) systems. This integration introduces a secure relay selection approach where secondary relays (SRs) join the network via an auction model. Virtual wallets are employed to manage balance within this model. The effectiveness of the framework in improving security is emphasized by the trustworthiness assessment mechanisms and offline blockchain frameworks.
The thesis delves into improving security in cooperative communication within vehicular wireless networks through the implementation of an intelligent spatiotemporal diversification moving target defense (MTD) scheme. This involves the strategic design of benign random data injection patterns to increase the attacker’s uncertainty and thus improve system security. Leveraging deep reinforcement learning (DRL), the optimization of vehicular relay configurations and data injection patterns is achieved, effectively improving the security of the system against eavesdropping attacks. However, vehicle mobility often leads to decisions based on outdated channel state information (CSI), impacting system performance. To address this, an adaptive, intelligent, and predictive relay selection scheme is crucial to ensure satisfactory performance despite outdated CSI. Motivated by this, we have leveraged the benefits of autoregressive models as well as the MTD to enhance the system adaptability and performance of a cooperative vehicular communication scenario with outdated CSI. Simulation results show significant improvements compared to the presented benchmark relay selection approaches. To optimize resource utilization and extend communication coverage, the thesis focuses on PLS in the unmanned aerial vehicle (UAV) domain. Specifically, a diversification-based artificial noise (AN) injection strategy is developed in a UAV-based network, which aims to confuse potential attackers while preserving the integrity of legitimate communications.
The main objective of the proposed framework is to maximize the average secrecy rate (ASR) by jointly optimizing UAV trajectory, transmit power, and splitting factor. Numerical results show the superiority of the proposed algorithm over benchmarks in terms of ASR and intercept probability, demonstrating its effectiveness in improving the security of UAV-based communications.
Finally, in the area of wireless communication technologies, this thesis delves into the convergence with extended reality (XR), presenting a security-aware cross-layer communication management framework for XR wireless communication. The framework prioritizes security without compromising latency, quality of experience (QoE), or resource allocation. Leveraging learning-based algorithms, especially DRL and real-time monitoring, this framework dynamically reprograms network configurations. In extensive simulations, the proposed algorithm shows superior performance, meeting the stringent security and QoE requirements for XR users.

Date of Award	21 May 2024
Original language	American English
Supervisor	Sami Muhaidat (Supervisor)