Measurements Clustering for Robustness Improvement of Indoor WLAN Propagation Models

For indoor Wireless Local Area Networks (WLANs) planning, an accurate propagation modelling is required. Semi-empirical models represent an efficient approach to the indoor channel coverage prediction. Their parameters are estimated from measurement results. In the case called ill-conditioned, the Least Square (LS) regression leads to a bad estimation of the model parameters and thus to numerical instability. In this paper, we present an approach based on the K-means clustering method which allows us to increase the estimation stability. Clustering is used to select tuning points for having a robust estimation. Index Terms – K-means, Radio Channel Modelling, Indoor Propagation, WLAN.