Locus: A Tool for Tag-less Location Sensing on 802.11 Wireless LANs with Display in SVG

Ali Taheri, Arvinder Singh and Emmanuel Agu

IEEE 802.11 Wireless Local Area Networks (WLANs) have become widely deployed and are fuelling a wide range of location-aware computing applications. Accurate user location information enables a wide range of location-dependent applications. Users may request to print to the closest printer, tourists may receive rich context information about their current location and in geocasting, network packets may be routed to users at a specified location.

GPS is currently the de facto standard for location sensing in outdoor wireless environments. However, GPS does not work well in indoor WLAN environments. Moreover, GPS requires dedicated hardware. Location sensing on 802.11 networks using commodity NICs and no additional tags or hardware is attractive due to its reduced cost, easier deployment and use, and improved indoor functionality.

Location fingerprinting is an increasingly popular location sensing technique which involves a two-phase process. First, received signal strength values (or radio map) from all Access Points (APs) at selected locations in a building are recorded during an offline calibration phase. Then, during the online phase, proximity-based matching algorithms are used to infer a user's location by comparing the current observed signal strength value to the pre-recorded values in the radio map database.

We present Locus, a software-only platform independent tool for location sensing on 802.11 wireless LANs with display in SVG. Locus was designed using object-oriented techniques and implemented in Java. Specific platform-independent modules are clearly abstracted and wrapped in Java Native Interface (JNI) classes. Specifically, we integrated WRAPI, a third party library for retrieving signal strength into our framework using JNI. A simple euclidean-distance-based proximity matching algorithm is used to infer user locations in real time.

User location information was displayed on actual building floor plans using SVG. Original building drawings were converted from the AutoCAD .dwg format to SVG and then used for displaying location. The Batik SVG development suite was used for developing our SVG interface. Basic graphical user interface capabilities like panning and zooming were added to improve usability. The SVG display area implements a multi-layer approach in which the building map is displayed in the bottom layer while users and various non-static resources such as printers are stored on distinct transparent layers. The composite SVG image is then generated at runtime using java calls to the Batik toolkit.

Locus was adequately tested in the Worcester Polytechnic Institute (WPI) library building. We are currently using Locus as a foundation to develop a wide range of location-aware applications and modules within our ubiquitous computing university campus environment. We also plan to release Locus to the research community.

This page is valid XHTML 1.1.