9 Million Bicycles? Extending Induction Loops with Bluetooth Sensing

Smart urban spaces need traffic information beyond traditional vehicular traffic. Detailed data about bicycle traffic in a city is highly valuable to adapt traffic lights, plan traffic routes, or provide information about situational travel times. However, traditional induction loops do not work well for bikes, and they cannot give information about routes and travel times. This paper shows how infrastructure data (induction loops) and data from Bluetooth sensing can be fused to derive better information about bicycle traffic in a smart city. We present a novel approach to dynamically determine Bluetooth ratios of different traffic participants based on rare events, and we evaluate the approach with data from a real-world study with 97 registered users, 23,074 Bluetooth devices and more than 174,917 Bluetooth detections over one week in the City of Oldenburg.