K-NN based positioning performance estimation for fingerprinting localization

The positioning system have been paid lots of attention to deal with numerous demands of providing location based services to users. To develop a positioning system, a target accuracy should be defined for reliable services. Though a same system is applied, however, the positioning performance is not always identical since signal environments and structure of indoor areas are different. If a system cannot achieve the target accuracy, the system and environments should be re-examined to find out a solution to meet the accuracy. As a result, additional costs are required to establish a positioning system soundly operable in broad areas. To reduce the cost, in this paper, a system for positioning performance estimation is addressed, especially for fingerprinting positioning system. The proposed system measures signal distances between location fingerprints constituting a radio-map to evaluate the similarity between the signal patterns. Ambiguous patterns bring larger positioning errors when signal fluctuations are introduced; therefore, by calculating the similarity the performance it gives an evidence to estimate degree of positioning errors. The experimental results in subway stations in Seoul, Korea, verify the feasibility of the proposed system.