Modeling of Indoor Positioning Systems Based on Location Fingerprinting
Kamol Kaemarungsi and Prashant Krishnamurthy Telecommunications Program School of Information Science. University of Pittsburgh 135 N. Bellefield Avenue, Pittsburgh. Pennsylvani
Given a set of system parameters and radio propagation characteristics. which are the number of access points. the grid spacing. the path loss exponent. and standard deviation of RSS. the accuracy of a positioning system can he determined in terms of the prohabilily of returning the correct location with the model presented in this manuscript. The analysis results provide a guideline on parameters to consider in designing and deploying an indoor positioning system. The system does not require a large number of access points (no more than five) to improve the accuracy performance. The number of access points and their locations depend on the network infrastructure and the budget. The cost of IEEE 802.1 Ib based access points is rapidly falling and it may not be a problem to add one or two more to provide additional coverage for indoor positioning. The grid spacing is closely tied to the application requirements of the indoor positioning system. The grid spacing must he selected properly and cannot be too small to provide a fine granularity or accuracy as the precision may he poor. A larger grid spacing may provide a ktter precision. but may render the location information very coarse. A grid spacing of 1 meter is found to be a suitable value based on our model and simulations. The standard deviation of the sampled RSS vector should be low in order to provide a better performance and a larger path loss exponent will benefit the precision. However. the standard deviation of RSS and the path loss exponent are not controllable parameters.
They vary depending on the indoor environment and building malerial. Larger standard deviations are reported inside large and open space buildings. while smaller standard deviations are found mostly in small and closed spaces [71. The path loss exponent also depends on the material inside buildings. A system designer should balance all of these system parameters by adjusting the number of access points and grid spacing in order to deliver a satisfactory accuracy and precision. In our ongoing work. we are trying to address the limitations of our model discussed previously and validate our model with real measurements.
