A Radio-Map clustering Algorithm for RSS based Localization using Directional Antennas

Position determination of nodes in wireless networks is becoming a prerequisite in industrial environments to fulfill various application and network needs ranging from device localization to safety or real-time behavior. Received Signal Strength (RSS) based localization techniques have been a research topic for many years, yet suffer from various physical effects. Usage of directed antennas could increase the localization performance and suppress disturbing effects, yet dramatically increase the complexity of localization especially in large scale networks. In this paper we propose a machine-learning aided approach for clustering the RSS map that significantly optimizes the position estimation of nodes in environments with antennas having arbitrary antenna patterns.