| The primary objective of this thesis is to seek a ground-based location scheme suitable for mobile positioning in cellular phone networks. To this end, several techniques are proposed to handle issues that may occur in a cellular network and which may deteriorate wireless location performance. These issues include the lack of signal availability due to co-channel interference, the inefficiency in mobile location calculations, and the significant Non-Line-of-Sight (NLOS) errors resulting from multipath propagation. With the IS-95 CDMA pilot signal as an example, signal availability---or hearability---is thoroughly analyzed. The analysis shows that hearability is poor for location purposes. To improve signal hearability, two methods that are known to be effective---the enhanced signal processing method and the idle period down link (lPDL) method---are fully discussed. Another promising solution in poor signal hearability environments is the combination of cellular network-based methods with other positioning methods. As an example, the integration of GPS and a cellular network is proposed. Better location performance can be obtained by epoch-by-epoch Least Squares (LS)-based integration schemes or by Kalman filter-based integration schemes. The position of a mobile handset is normally obtained by solving non-linear equations. However, it represents a high computational burden and may suffer from a convergence problem. To resolve these issues, an enhanced two-step LS solution is proposed for hybrid time difference of arrival (TDOA)/angle of arrival (AOA) wireless location schemes. This method can provide performance which is almost equivalent to that of Taylor-series-based solutions while imposing a low computational burden. Because NLOS errors within time of arrival (TOA), TDOA, and AOA measurements are very large compared to receiver noise, NLOS errors should be mitigated before the measurements are used in the position calculation. Two NLOS error mitigation methods are proposed. One is a distribution function-based method which depends on system redundancy and a high clear intersection density. The other is a channel estimation-based method which mitigates NLOS errors by using only the earliest signal from among all multipath replicas. The effectiveness of all of the proposed methods has been proved by simulation tests, verifying that these methods can be successfully applied in an actual wireless location system design. |
