Self-adaptive Wi-Fi indoor positioning model

Abstract

Wi-Fi based indoor positioning, which is based on attenuation of Received Signal Strength Indicator (RSSI) is an emerging Location Based Service (LBS) technology. As positioning accuracy is sensitive to environmental factors, most of the existing algorithms based on experimental test perform badly without adaptation to dynamics of environment. In this paper, we propose an indoor positioning method by locating the representation of a cluster within similar environments. The K-Means algorithm is used to extract the similarities of the objects within the nearby area. To overcome the problem of parameter determination under the circumstances of lack of fingerprint and extra hardware, we proposed a Log-normal shadowing model (LNSM) with Artificial Neural Networks to estimate distance enabling the parameters to be dynamically adjusted according to the change of the environment. The experimental results of one day auto fair data demonstrate the performance of our method with a higher degree of accuracy than other methods.