Towards the Next Generation of Location-Aware Communications

Abstract

This thesis is motivated by the expected implementation of the next generation mobile networks (5G) from 2020, which is being designed with a radical paradigm shift towards millimeter-wave technology (mmWave). Operating in 30--300 GHz frequency band (1--10 mm wavelengths), massive antenna arrays that provide a high angular resolution, while being packed on a small area will be used. Moreover, since the abundant mmWave spectrum is barely occupied, large bandwidth allocation is possible and will enable low-error time estimation. With this high spatiotemporal resolution, mmWave technology readily lends itself to extremely accurate localization that can be harnessed in the network design and optimization, as well as utilized in many modern applications. Localization in 5G is still in early stages, and very little is known about its performance and feasibility.

In this thesis, we contribute to the understanding of 5G mmWave localization by focusing on challenges pertaining to this emerging technology. Towards that, we start by considering a conventional cellular system and propose a positioning method under outdoor LOS/NLOS conditions that, although approaches the Cram\'er-Rao lower bound (CRLB), provides accuracy in the order of meters. This shows that conventional systems have limited range of location-aware applications. Next, we focus on mmWave localization in three stages. Firstly, we tackle the initial access (IA) problem, whereby user equipment (UE) attempts to establish a link with a base station (BS). The challenge in this problem stems from the high directivity of mmWave. We investigate two beamforming schemes: directional and random. Subsequently, we address 3D localization beyond IA phase. Devices nowadays have higher computational capabilities and may perform localization in the downlink. However, beamforming on the UE side is sensitive to the device orientation. Thus, we study localization in both the uplink and downlink under multipath propagation and derive the position (PEB) and orientation error bounds (OEB). We also investigate the impact of the number of antennas and the number of beams on these bounds. Finally, the above components assume that the system is synchronized. However, synchronization in communication systems is not usually tight enough for localization. Therefore, we study two-way localization as a means to alleviate the synchronization requirement and investigate two protocols: distributed (DLP) and centralized (CLP).

Our results show that random-phase beamforming is more appropriate IA approach in the studied scenarios. We also observe that the uplink and downlink are not equivalent, in that the error bounds scale differently with the number of antennas, and that uplink localization is sensitive to the UE orientation, while downlink is not. Furthermore, we find that NLOS paths generally boost localization. The investigation of the two-way protocols shows that CLP outperforms DLP by a significant margin. We also observe that mmWave localization is mainly limited by angular rather than temporal estimation.

In conclusion, we show that mmWave systems are capable of localizing a UE with sub-meter position error, and sub-degree orientation error, which asserts that mmWave will play a central role in communication network optimization and unlock opportunities that were not available in the previous generation.