Physical layer security in cellular networks

Abstract

In this work, we provide a theoretical study of physical layer security (PHYLS) performance in full-duplex (FD) small-cell networks. Here, the base stations (BSs) and user equipments (UEs) follow from the homogeneous Poisson point process (PPP)-based abstraction model. First we consider the network where BSs serve a single user in one resource block where PHYLS in both scenarios of HD and FD UEs are analysed and compared. We derive the ergodic secrecy rate in both downlink (DL) and uplink (UL) in presence of a field of passive eavesdroppers (EDs). To facilitate FD communications, we take into account (i) successive interference cancellation (SIC) capability at the UE side via guard regions of arbitrary radii, and (ii) residual self-interference (SI) at the BS side using Rician fading distribution with arbitrary statistics. We study the affect of density of BSs and EDs in the FD mode and compare it to its HD counterpart. Next, we consider the case where BSs serve several HD single-antenna users at the same time and frequency with assistance of multiple antennas. We investigate the small-cell network PHYLS performance in the presence of a Poisson field of EDs, under the different scenarios of passive and colluding eavesdropping. Considering linear zero-forcing (ZF) beam-forming in the multiple-input multiple-output (MIMO) scenario, we characterize the UL and DL ergodic secrecy rates and derive closed-form expressions for the different useful and interference signals statistics. Moreover, we look into the impact of SIC and SI capabilities in the UE and BS side, respectively. Thereafter, we derive explicit expressions of the key performance indicators using tools from machine learning. In particular, for certain special cases of interest, we apply supervised learning-based non-linear curve-fitting techniques to large sets of (exact) theoretical data in order to obtain closed-form approximations for the different ergodic rates and ergodic secrecy rates under consideration. Our findings, obtained from theoretical analysis and system-level simulations, indicate that the FD functionality, in addition to enhancing the spectral efficiency (SE), can significantly improve the PHYLS performance, especially with the aid of multi-antenna communications and interference cancellation schemes. Finally, we highlight several promising future research directions in relation to the outcomes of this work, including the application of physical layer security and full-duplex operation in the context of 5G networks and beyond.

Date of Award	1 Dec 2019
Original language	English
Awarding Institution	King's College London
Supervisor	Abdol-Hamid Aghvami (Supervisor)