Estimating Single and Multiple Target Locations Using K-Means Clustering with Radio Tomographic Imaging in Wireless Sensor Networks

Geolocation involves using data from a sensor network to assess and estimate the location of a moving or stationary target. Received Signal Strength (RSS), Angle of Arrival (AoA), and/or Time Difference of Arrival (TDoA) measurements can be used to estimate target location in sensor networks. Radio Tomographic Imaging (RTI) is an emerging Device-Free Localization (DFL) concept that utilizes the RSS values of a Wireless Sensor Network (WSN) to geolocate stationary or moving target(s). The WSN is set up around the Area of Interest (AoI) and the target of interest, which can be a person or object. The target inside the AoI creates a shadowing loss between each link being obstructed by the target. This research focuses on position estimation of single and multiple targets inside a RTI network. This research applies K-means clustering to localize one or more targets. K-means clustering is an algorithm that has been used in data mining applications such as machine learning applications, pattern recognition, hyper-spectral imagery, artificial intelligence, crowd analysis, and Multiple Target Tracking (MTT)