Towards Localisation in Next-generation Wireless Systems

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)

Abstract

Localisation in wireless communication systems is an important topic that has many applications. Accurate geolocation in urban environments is a well understood challenge and Global Navigation Satellite System (GNSS) based technologies like GPS yield poor accuracy in non-line-of-sight (NLOS) scenarios. Use of multiple GNSS systems improves location accuracy but is still affected by urban canyons. Mobile network-based schemes offer an alternative, particularly when antenna arrays are deployed. Massive antenna arrays in Massive Multiple-Input-Multiple-Output (Massive MIMO) present an opportunity because of the possibility to use inexpensive, low-power and low-precision components with greatly reduced complexity/cost, in addition to Massive MIMO being a core component of 5G.

This thesis reviews the problem of localisation in urban environments and investigates the techniques that are relevant towards achieving localisation in next generation wireless systems. It evaluates the NLOS problem and proposes schemes that use machine leaning techniques in the form of Least-Squares Support Vector Machines (LSSVMs), to address the challenges. It also investigates Direction of Arrival (DOA) estimation, which is a step towards localisation, using the Bristol Massive MIMO testbed.

The proposed location specific approach presented in this thesis is a new framework which is shown to achieve a best case NLOS identification accuracy of greater than 98 percent. This result exceeds reported accuracies for existing NLOS identification techniques. Another proposed direct localisation approach achieves an 80th percentile probability location accuracy of 10 metres without utilising NLOS identification and mitigation. DOA estimation experiments demonstrate the possibility of performing estimation using subsets of antennas on a single base station antenna array. An outdoor experiment demonstrated a best case DOA RMSE of 2 degrees achieved for 3 different User Equipment (UE) positions.

Overall, the results in this thesis demonstrate techniques that can be employed to solve the intermediate challenges towards device localisation and can contribute to the design and operation of next-generation systems. Significant benefits for mobile wireless systems can be derived from localisation. New handover strategies, new dynamic resource, power and pilot allocation schemes that take advantage of the location information, can also be developed.
Date of Award24 Jun 2021
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorSimon M D Armour (Supervisor) & Mark A Beach (Supervisor)

Keywords

  • Localisation
  • 5G
  • MIMO
  • Support vector machine
  • TDOA
  • DOA

Cite this

'