Sensor network topology estimation using time-series data from infrared human presence sensors

We describe a method for accurately estimating the topology of sensor networks from time-series data collected from infrared proximity sensors. Our method is a hybrid combining two different methodologies: ant colony optimization (ACO), which is an evolutionary computation algorithm; and an adjacency score, which is a novel statistical measure based on heuristic knowledge. We show that, using actual data gathered from a real-world environment, our method can estimate a sensor network topology whose accuracy is approximately 95% in our environment. This is an acceptable result for real-world sensor-network applications.