Physical Layer Security and Latency Optimization for Internet of Things Communications

Abstract

The Internet of Things (IoT) aims to connect all things to the internet to facilitate intelligent identifying, locating, tracking, monitoring, and managing of objects and information. The IoT paradigm will be supported by advanced wireless communications technologies and fifth-generation (5G) cellular networks. As such, new network architectures and protocols should be developed to ensure successful deployments of IoT communications systems in many domains including smart homes, transportation, logistics, smart grids and healthcare. In order to extend existing wireless network coverage and support the robustness and connectivity of IoT, unmanned aerial vehicle (UAV) has been proposed as an ideal IoT platform due to its flexibility and mobility. Moreover, mobile edge computing (MEC) which provides considerable computing resource at the edge of the network to support IoT applications and services in real time is another key enabling technology for IoT. However, realizing the vision of IoT is challenging due to many difficulties that need to be addressed. Particularly, physical layer security (PLS) and low-latency design are the two main challenges. On the one hand, due to the openness of wireless transmission medium, the communications between the legitimate transmitter-receiver pair can be readily attacked by eavesdroppers or spoofers where the confidentiality and integrity of the information cannot be guaranteed. On the other hand, with the tremendous amount and variety of data to be collected and processed in IoT networks, achieving low-latency communications is challenging and has become one of the major bottlenecks of IoT. In this thesis, we develop new efficient algorithms to improve the PLS and latency in IoT communications based on UAV and MEC technologies to address the aforementioned challenges of IoT. We first develop a secure UAV-enabled communication framework by exploiting a UAV-based friendly jammer which emits artificial noise to prevent eavesdropping attacks on legitimate ground nodes when the eavesdropper location is unknown. We propose an efficient algorithm to maximize the security performance by jointly optimizing the 3D deployment and jamming power of the UAV jammer. Numerical results show that our proposed iterative algorithm performs close to an exhaustive search with significantly reduced complexity. Next, we propose a more general IoT communications model where both the latency and PLS performance are jointly considered in an MEC network. We formulate a latency minimization problem by jointly optimizing the user's transmit power, computing capacity allocation, and user association subject to PLS performance and computing resource constraints. Numerical results show that our proposed solution outperforms baseline strategies over a wide range of computing capacities and highlight a fundamental tradeoff between latency and security in an MEC network. Subsequently, we extend the latency-security tradeoff analysis to a UAV-enabled MEC network, where multiple ground users offload large computing tasks to a nearby legitimate UAV in the presence of multiple eavesdropping UAVs with imperfect locations. For this system, we design a low-complexity iterative algorithm to maximize the minimum secrecy capacity subject to latency, minimum offloading and total power constraints. Numerical results show that our proposed algorithm significantly outperforms baseline strategies over a wide range of UAV self-interference (SI) efficiencies, locations and packet sizes of ground users. Furthermore, we show that there exists a fundamental tradeoff between the security and latency of UAV-enabled MEC systems. Finally, to protect wireless communications from potential spoofing attacks, we investigate received signal strength (RSS)-based physical layer authentication (PLA) in UAV-enabled communication systems based on game theory. We first model an authentication hypothesis test based on the RSS distance and derive the false alarm rate and miss detection rate. We then formulate a zero-sum PLA game to model the interactions between the spoofer and the UAV receiver. The Nash equilibrium (NE) and its existence condition for the proposed PLA game are also derived. Monte Carlo simulation results accurately verify our analytical expressions for the false alarm rate and miss detection rate.