Techniques for Mobile Location Estimation in UMTS

Abstract

The subject area of this thesis is the locating of mobile users using the future 3rd generation spread spectrum communication system UMTS. The motivation behind this work is twofold: firstly the United States Federal Communications Commission (FCC) mandated the provision of user location into services in the United States of America due to the increasing number of emergency calls originating from unknown locations. Secondly the user location can enable a number of other potentially profit–making applications and services. These are generally thought to be the important new applications of the third generation mobile networks. The UMTS standard has now made provision for a time difference of arrival based mobile user location system in which the mobile measures time differences of arrival of received signals from surrounding base stations (BS’s). There are two main problems to such a technique: firstly the problem of detecting enough base stations to make a location fix, the so called ‘hearability’ problem. In spread spectrum systems all base stations transmit on the same bandwidth thus non–serving BS’s may not be detectable in normal operation. The second problem is non–line of sight (NLOS) propagation, in which time difference measurements (or any other measurement types) may be corrupted significantly, thus causing significant location error. The thesis of this work is that these two problems can be entirely overcome using spatial filtering of measurements and location estimates. Two constraints that are placed on the filtering algorithms are that the operation should be real time and that the precise distribution of NLOS errors is unknown (though certain key characteristics are exploited). A channel model is first developed, which specifically characterises line of sight and NLOS transitions as well as out of cell radio wave propagation. Several scenarios are then simulated. Slow moving users, low hearability and heavily NLOS conditions pose the biggest challenge. Spatial filtering is achieved by Kalman filters adapted to the problem, as well as simple averaging filters. Results show that improved location accuracy (to within FCC recommendations) is possible in all considered scenarios with spatial filtering as well as improved robustness to low hearability. The detection stage of the receiver is also analysed in detail and methods to improve hearability are presented. The performance of a hybrid location system using angle of arrival measurements of the mobile at the serving BS is also assessed. A fairly pessimistic model for the spread of NLOS errors is used, however significant location improvement is noted in several scenarios. Worst performance occurs in urban scenarios so finally a novel approach to user location is described which is robust to NLOS propagation conditions and also overcomes the hearability problem since only measurements at the serving BS are required. The technique, termed Scatterer Back Tracing (SBT), uses and requires multipaths to calculate the mobile location. Results suggest this SBT can provide extremely high location accuracy but is very sensitive to measurement noise.