No abstract

This paper illustrates two techniques for improved estimation of the location of Mobile Stations (MS) in cellularNetworks. The first approach is the statistical method in which signal properties are treated as random variables which are statistically dependent on the location of the transmitter and the receiver. Location estimation for a set of observed signal strengths at a specific location is done as an inference problem. In the second approach, Database Correlation, signal information seen by an MS is stored in the form of fingerprints. To estimate the location, difference between the measurement and the database fingerprints is calculated using a correlation algorithm. Database fingerprints with least difference are selected as the nearest fingerprints for the location to be estimated. The two techniques have been tested through extensive measurements in two environments, urban and sub urban, to verify their performances. Two different implementations have been developed and presented for the statistical technique examined through simulations in the literature. In the implementation of database correlation method, different correlation algorithms and measurement conditions are proposed, implemented and tested. In addition, the accuracies of two techniques have been presented compared to the simple geometrical method for position estimation.

No abstract

No abstract