Online event clustering in temporal dimension

摘要: This work is motivated by a real-life application that exploits sensor data available from traffic light control systems currently deployed in many cities around the world. Each consists of an induction loop generates stream events triggered whenever metallic object e.g. car, bus, or bicycle, detected above sensor. Because red phase lights objects are usually divided into groups move together. Detecting these as long they pass through useful for estimating status toad networks such car queue length detecting anomalies. In this work, given contains observations event, detection moving object, together with timestamps indicating when happen, we study problem clusters real-time based on proximity event's occurrence time. We propose efficient algorithm scales up to large streams extracted thousands sensors city London. Moreover, our better than baseline algorithms terms clustering accuracy. demonstrate motivations showing use-case which results used lengths road and