Mobile Crowdsensing Based Road Surface Monitoring Using Smartphone Vibration Sensor and Lorawan

Abstract

Road surface monitoring is a critical activity in road transport infrastructure management. In this paper, we present a mobile crowd-sensing based road surface monitoring using Smartphone sensors and a LoRaWAN network. Using the accelerometer and GPS sensors of the Smartphone, it's possible to measure vibration and where it happens, enabling the generation of reports of road conditions and anomalies. These reports can be transmitted by low-cost, low-power and secure communication links provided by the LoRaWAN network infrastructure thus saving the added cost of transmitting them over the cellular network. We focus on monitoring the asphalt road surface using a machine learning model classifying the vibration generated by vehicles as pothole, speed bump, damaged road or patched road. As proof of concept, we developed a mobile application with a built-in machine learning model to detect and classify road condition. To reduce the bandwidth consumption, the application reports only road condition classification instead of sending the raw vibration signal. The main objective is to reduce the burden of manual inspection and measurement while minimizing communication cost. Our approach was tested and evaluated by real-world experiments in a road segment.