Accuracy Enhancement with Integrated Database Construction for Indoor WLAN Localization

Abstract

In this paper, we rely on the neighborhood relations of the physically adjacent Reference Points (RPs) to construct a physical neighborhood database with the purpose of enhancing the accuracy of the Receive Signal Strength (RSS) fingerprint based localization algorithms in Wireless Local Area Network (WLAN) environment. First of all, based on the Most Adjacent Points (MAPs) and their corresponding Physically Adjacent Points (PAPs), we construct the Feature Groups (FGs), and then calculate the New Reference Point (NRP) with respect to each FG. Second, the RSS at each NRP is estimated by using the least square method based surface interpolation algorithm. Finally, we apply the K Nearest Neighbor (KNN), Weighted KNN (WKNN), and Bayesian inference algorithms to locate the target. The experimental results show that the proposed integrated database construction helps a lot in improving the localization accuracy of the widely-used KNN, WKNN, and Bayesian inference algorithms.