LOCATOR: location estimation system For wireless LANs

Abstract

With the wide spread growth of mobile computing devices and local area wireless networks, wireless network providers have started to target the users with value-added services based on the users' location information. Thus, location awareness and user tracking in indoor wireless networks has become an increasingly important issue. To address the issue, we have developed LOCATOR, a radio frequency based system for location estimation of users in indoor wireless networks. Our system works by building a radio map of the network site, which involves taking signal strength samples at various points in the wireless network, and then using this radio map, the system estimates the user's current location from the value of his current observed signal strength. Although the use of an RF-based technique for location estimation is not new, but LOCATOR is unique in the way it builds and manages the radio map to process location queries. Moreover, in addition to using ideas of probability to model the problem, we have devised a multi-level clustering based algorithm that work in conjunction with an interpolation scheme for a more efficient and accurate location estimation. We have tested our system on 802.11b wireless network testbeds and have been able to achieve an accuracy of location estimation to within 4 feet of the actual user location with 90% probabilit.