Crowdsourced Bicycle Monitoring and Finding With Riding User Recognition Based on Internet of Things Technologies

In this article, we propose a crowdsourced bicycle monitoring and finding (CBMF) framework with riding user recognition for rider safety and bike security based on Internet of Things (IoT) technologies. According to our review of relevant research, CBMF is the first framework that provides the following features: 1) it can use the inertial data to determine the normal/dangerous state of the bicycle; 2) it can recognize whether the rider is the owner based on the inertial data of cycling; and 3) if the bicycle equipped with the IoT device is stolen/missing, the current locations of the stolen bicycle can be reported by surrounding smartphones that receive the iBeacon signals with an abnormal bit proactively broadcasted by the equipped IoT device. In particular, the densely connected convolutional network is explored in CBMF to extract diverse features of riding for improving rider recognition accuracy. An Android-based prototype with the Jetson Nano developing board is implemented to verify the feasibility and superiority of our framework. Experimental results show that our developed deep learning model outperforms the existing methods and can accurately recognize the riding user (with over 90% accuracy). Moreover, the crowdsourced sensing approach can precisely locate the position of the bicycle (reducing about 50% of positioning errors), which increases the possibility of getting back the stolen/missing bicycle.