Sensor similarity search in the Web of Things

An increasing number of sensors is being connected to the Internet and their output is published on the Web, resulting in the formation of a Web of Things (WoT) that will soon connect tens of Billions of devices. We propose sensor similarity search, where given a sensor, other sensors on the WoT are found that produced similar output in the past. At the heart of our approach is an algorithm that exploits fuzzy sets for efficiently computing a similarity score for a pair of sensors that is used to obtain a ranked list of matching sensors. Using sensor data sets from real deployments, we find that this approach results in a high accuracy.