An efficient cluster-based outdoor user positioning using LTE and WLAN signal strengths

In this paper we propose a novel cluster-based RF fingerprinting method for outdoor user-equipment (UE) positioning using both LTE and WLAN signals. It uses a simple cost effective agglomerative hierarchical clustering with Davies-Bouldin criterion to select the optimal cluster number. The positioning method does not require training signature formation prior to UE position estimation phase. It is capable of reducing the search space for clustering operation by using LTE cell-ID searching criteria. This enables the method to estimate UE positioning in short time with less computational expense. To validate the cluster-based positioning real-time field measurements were collected using readily available cellular mobile handset equipped with Nemo Handy software. Output results of the proposed method were compared with a single grid-cell layout based RF fingerprinting method. Simulation results show that if a single LTE and six WLAN signal strengths are used then the proposed method can improve positioning accuracy of 35% over the grid-based RF fingerprinting.