Federal Communication Commission (FCC) regulations state that by October 1, 2001, wireless callers to E-911 must be located within a radius of 125 m of their actual locations at least 67% of the time. While there are many different wireless location-finding techniques, not all of them meet the FCC requirements. One location-finding method, the multipath fingerprint method, has been proposed, but is neither well known, nor well explored. The multipath fingerprint method is a network-based wireless E-911 location-finding technique. It matches fingerprints extracted from a caller's signal with a database of previously created fingerprints associated with known caller positions. Fingerprints are created from time of arrival (TOA), angle of arrival (AOA), and signal strength parameters extracted from input signals.; In this dissertation, I demonstrate the feasibility of the multipath fingerprint method for locating wireless callers in an urban environment by using extensive electromagnetic simulation and limited measurement data. I show that from a theoretical standpoint, a city can be classified by unique fingerprints, and that this method has the potential to meet and beat the FCC requirements.; The multipath fingerprint system is composed of three parts—the fingerprint database, the feature extractor, and the location classifier. All three parts are examined in this dissertation based on simulation data generated using the ray-tracing code CPATCH, which is based on the Shooting and Bouncing Ray (SBR) technique. First, a fingerprint database is created for a symmetrical four-by-four building Computer Aided Design (CAD) model using CPATCH. Fingerprint characteristics are studied in detail, including the effects of bandwidth, array size, and traffic. A location classifier is created using simulated data from a CAD model of downtown Austin, Texas, for two-dimensional results (time and angle of arrival). One-dimensional classification results are also shown. The feature extractor is developed using a proposed two-dimensional MUSIC (MUltiple SIgnal Classification) algorithm, which calculates time and angle of arrival for a circular array with subspace smoothing. This algorithm extends the one-dimensional Angle of Arrival MUSIC algorithm for circular arrays and integrates one-dimensional subspace smoothing using spatial signatures. Two-dimensional fingerprints are created using MUSIC-extracted data and compared to CPATCH-generated fingerprints.
展开▼
