Indoor WLAN positioning Using Hybrid SVM Hyperplane Margin Clustering and Regression

This paper proposes a novel indoor Wireless Local Area Network (WLAN) positioning algorithm by using the Support Vector Machine (SVM) Hyperplane Margin Clustering and Regression (SVMCR). First of all, we rely on the SVM Hyperplane Margin Clustering (SVMC) to reduce the search space of the fingerprint database. Second, we use the Support Vector Regression (SVR) to characterize the relations of the Received Signal Strengths (RSSs) and physical locations for the sake of achieving the accurate positioning. Compared with the conventional indoor WLAN positioning algorithms, the proposed one significantly reduces the storage overhead, as well as guarantees the high positioning accuracy.