The influence of k-NN parameters on the localization accuracy of WSNs in indoor environment

Abstract

Localization is one of the most complex issues in wireless sensor networks (WSN) domain, particularly, when the network is in indoor environments where the effect of multipath propagation is more pronounced. In this paper we have focused on several factors that effect on the localization accuracy, including the target environment itself, as well as the various parameters of the program proposed in this paper, which is basically based on fingerprinting method, which uses the k nearest neighbor's technique (k-NN) to estimate target positions. The parameters discussed is the number k of neighbors around the target, the choice of the method to calculate the distance between the target and the various neighboring (Euclidean or Manhattan distance), and Wt the weight function.