A number of techniques for indoor and outdoor WiFi localization using received signal strength (RSS) signatures have been published. Little work has been performed to characterize the RSS signatures used by these WiFi localization techniques or to assess the accuracy of current channel models to represent the signatures. Without accurate characterization and models of the RSS signatures, a large amount of empirical data is needed to evaluate the performance of the WiFi localization techniques. The goal of this research is to characterize the RSS signatures, propose channel model improvements based on the characterization, and study the performance of channel models for use in WiFi localization simulations to eliminate the need for large amounts of empirical data measurements. In this thesis, we present our empirical database of RSS signatures measured on the Worcester Polytechnic Institute campus. We use the empirical database to characterize the RSS signatures used in WiFi localization, showing that they are composed of connective segments and influenced by the access point (AP) location within a building. From the characterization, we propose improving existing channel models by building partitioning the signal path-loss using site-specific information from Google Earth. We then evaluate the performance of the existing channel models and the building partitioned models against the empirical data. The results show that using site-specific information to building partition the signal path-loss a tighter fit to the empirical RSS signatures can be achieved.