MobisenseCar: A Mobile Crowd-Based Architecture for Data Acquisition and Processing in Vehicle-Based Sensing

The use of wireless technology via smartphone allows designing smartphone applications based on OBD-II for increasing environment sensing. However, uploading of vehicle’s diagnostics data via car driver’s tethered smart phone attests a long Internet latency when a large number of concurrent users use the remote mobile crowdsensing server application simultaneously, which increases the communication cost. The large volume of data would also challenge the traditional data processing framework. This paper studies design functionalities of mobile crowdsensing architecture applied to vehicle-based sensing for handling a huge amount of sensor data collected by those vehicle-based sensors equipped with a smart device connected to the OBD-II interface. The proposed MobiSenseCar uses Node.js, a web server architecture based on single-thread event loop approach and Apache Hive platform responsible for analyzing vehicle’s engine data. The Node.JS is 40% faster than the traditional web server side features thread-based approach. Experiment results show that MapReduce algorithm is highly scalable and optimized for distributed computing. With this mobile crowdsensing architecture it was possible to monitor information of car’s diagnostic system condition in real time, improving driving ability and protect the environment by reducing vehicle emissions.