Minimizing Calibration Efforts for an Indoor 802.11 Device Location Measurement System
John Krumm and John Platt
In: Machine Learning Meets the User Interface, 12 December 2003, Whistler, Canada.
Using an 802.11 wireless client as a location sensor is an increasingly popular way of enabling location-based services. Triangulation on signal strengths from multiple access points can be used to pinpoint location down to a few meters. However, this level of accuracy comes at the price of a manual, tedious, spatially high-density calibration of signal strength as a function of location. This paper presents a new 802.11 location algorithm based on a relatively coarse calibration. This helps answer the question of how accurate location can be computed based on a realistic level of calibration effort. The algorithm uses an interpolation function that gives location as a function of signal strength. As such, it is suited to maintaining some degree of performance in spite of reduced calibration data. We use this feature to test the effect of reducing the number of calibration readings per location and the number of locations visited during calibration. Our experiments show that calibration effort can be significantly reduced with only a minor reduction in spatial accuracy. This effectively diminishes one of the most daunting practical barriers to wider adoption of this type of location measurement technique.