Enhanced weighted K-nearest neighbor algorithm for indoor Wi-Fi positioning systems

Location-based systems for indoor positioning have been studied widely owing to their application in various fields. The fingerprinting approach is often used in Wi-Fi positioning systems. The K-nearest-neighbor fingerprinting algorithm uses a fixed number of neighbors, which reduces positioning accuracy. Here, we propose a novel fingerprinting algorithm, the enhanced weighted K-nearest neighbor (EWKNN) algorithm, which improves accuracy by changing the number of considered neighbors. Experimental results show that the proposed algorithm gives higher accuracy.