Heinrich Schmitzberger: Harnessing Wireless LAN Communication for Scalable Indoor Localization and Tracking, Doctoral thesis, Department of Business Informatics - Software Engineering, Johannes Kepler University Linz, September 2012.



Localization of people is an integral aspect of mobile computing research that has been inves-tigated primarily on the basis of radio or sound sensor technologies such as Bluetooth, WLAN or ultrasound. Due to the physical behavior of these carrier technologies in different environments, a principal distinction between outdoor and indoor localization has evolved. As an outstanding example of applied localization technologies, the Global Positioning System demonstrates the feasibility of a client-based solution to the outdoor localization problem on a global scale. For indoor spaces on the other hand, no convincing counterpart has emerged so far. The problem of localizing people indoors is still considered a matter of intense research.
Basically, two approaches exist for determining the user’s location indoors. The client-based variant assumes that sensor functionality is integrated in the mobile device of the user. The infrastructure-based solution uses sensors embedded in the environment. Recently, client-based WLAN localization has been the prevalent choice of technology for both mobile application research and the consumer market mainly because of the high availability of WLAN communication hardware. The major drawback of this approach is attributable to the heterogeneity of client hardware configurations the systems have to deal with (e.g., diverse antenna and radio characteristics), which results in a decline of achievable system accuracy and latency. Furthermore, additional software is required at the client device for realizing sensor functionality. As this sensor software depends on platform- and WLAN chipset-specific capabilities, the amount of supported devices is limited.
This thesis presents the design, implementation and evaluation of an infrastructure-based in-door localization and tracking framework that allows service provisioning without any client prerequisites but an actively transmitting wireless LAN interface. For measuring the traffic emitted by the client, the system employs a network of stationary, passive WLAN sensors realized on the basis of low cost off-the-shelf access point hardware and modified with soft-ware specially developed for the purpose of sensing wireless LAN communication. With re-gard to the infrastructure-based variant of localization, scalability issues arise (due to e.g., localization algorithm complexity, growing reference databases, heavy network traffic) that have been a key impediment for the usage of such systems in real-life large scale setups in the past. This work investigates these issues and offers a solution on the basis of sensor-side pre-processing and dynamic measurement filtering (e.g., client whitelisting) that is able to per-form on a large scale while delivering latency and accuracy results suitable for the require-ments of mobile computing applications.