Analysis and Design of Multi-antenna Systems for Physical-layer Security and Interference Characterization

Abstract

It has become crucial to provide effective solutions to the security issues associated in a wireless transmission medium. The essential goal of physical-layer security is to enable the transmission of confidential messages over a wireless medium in the presence of unauthorized eavesdroppers. We begin with an overview of the foundations dating back to the pioneering work on physical-layer security, and the evolution of secure transmission strategies from point-to-point channels to multiple-antenna systems. The explosion of interest in multi-antenna systems led to the realization that exploiting the available spatial dimensions could also enhance the secrecy capabilities of wireless channels. Our work analyzes and optimizes multi-antenna systems under various assumptions on system model and channel information: transmission only versus joint transmission/jamming, perfect channel state information (CSI), imperfect CSI, and completely unknown CSI. With perfect CSI: We first study the problem of joint transmit and cooperative jamming to maximize the secrecy rate of MISOSE wiretap channels. We reduce the problem of maximizing the secrecy rate to the problem of finding the optimal jamming levels, and solve it by a one-dimensional search that is computationally affordable. We then study MIMOME wiretap channel with transmit beamforming (the rank-1 constraint on the transmit covariance matrix) and cooperative jamming. We propose an iterative algorithm by relating the MIMOME channel to the effective MISOSE channel. With imperfect CSI: assuming the CSI of the receiver and the eavesdropper channel belong to given uncertainty sets, we optimize the worst-case secrecy rate of the MISOSE wiretap channel. By proving the saddle-point solution, we transformed the worst-case optimization problem into a quasi-convex optimization problem. Moreover, for jamming-aided MISOSE channel with imperfect CSI, we propose to approximate the worst-case secrecy rate as the worst-case SINR-ratio problem. A minimax solution is shown to exist for the problem, based on which some insightful results of the optimal transmit and jamming covariances are obtained. A quasi-convex optimization algorithm that is tractable and efficient is provided for solving the worst-case SINR-ratio problem. With no CSI: We study how cooperative jamming helps improve the secure throughput of large decentralized networks where both the locations and CSI of eavesdroppers are unknown. The spatial distributions are modeled as Poisson point processes. The helping jammers broadcast artificial noise that confuses eavesdropper but zero-forcing to the legitimate receiver. A jamming protocol based on IEEE 802.11 RTS/CTS is proposed. Closed-form results analyze the benefits of jamming on secure communications. This thesis continues on analyzing large distributed networks using stochastic geometry. The geometry of the locations of the nodes plays a key role since it determines the signal-to-interference-plus-noise-ratio (SINR) at each receiver. The advantages of using stochastic geometry are: 1) performance metric can be exactly derived in some important cases, and tightly bounded in many others; 2) performance depends on fundamental network parameters, such as the densities of the underlying point processes. Design insights are obtainable from these performance expressions. Specifically, we study the performance of minimum-mean-square-error (MMSE) receivers under supposition of multiple Poisson point processes, non-homogeneous Poisson process and cluster Poisson point process of interferers. We discover that the SINR outage exhibits a superposition property for multiple homogeneous (or non-homogeneous) Poisson fields of interferers. Using this property, we extend the outage analysis to Poisson clustered processes which is formed by Poisson clusters consisting of Poisson distributed children points.