On the impact of node placement and profile point selection on indoor localization

Abstract

We present an indoor localization technique based on RF profiling using the received signal strength (RSS) measurements from a set of pre–selected reference points. We do not attach any interpretative significance to the measurements other than use them to calculate their difference from the measurements of the reference points. We study the performance of our technique in an environment with multiple adjacent rooms and find that it gives better results compared to the application of the k-Nearest Neighbor algorithm that has been used in the literature for the same task. We also study the proposed scheme and two other well–known localization schemes, with respect to the sensitivity of the localization on the number and layout of the reference points, as well as on the number and layout of the deployed fixed points (pegs) from where the measurements are collected. We find that one can achieve good localization performance with either fewer reference points or with fewer pegs as long as their layout is chosen carefully.