A Novel On-Demand Vehicular Sensing Framework for Traffic Condition Monitoring
With the increased need for mobility and the overcrowding of cities, the area of Intelligent Transportation aims at improving the efficiency, safety, and productivity of transportation systems by relying on communication and sensing technologies. One of the main challenges faced in Intelligent Transportation Systems (ITS) pertains to the real time collection of traffic and road related data, in a cost effective, efficient, and scalable manner. The current approaches still suffer from problems related to the mobile devices energy consumption and overhead in terms of communications and processing. To tackle the aforementioned challenges, we propose in this paper a novel infrastructure-less on-demand vehicular sensing framework that provides accurate road condition monitoring, while reducing the number of participating vehicles, energy consumption, and communication overhead. Our approach is adopting the concept of Mobile Sensing as a Service (MSaaS), in which mobile owners participate in the data collection activities and decide to offer the sensing capabilities of their phones as services to other users. Unlike existing approaches that rely on opportunistic continuous sensing from all available cars, this ability to offer sensory data to consumers on demand can bring significant benefits to ITS and can constitute an efficient and flexible solution to the problem of real-time traffic/road data collection. A combination of prototyping and traffic simulation traces are used to realize the system, and a variety of test cases are used to evaluate its performance. When compared to the traditional continuous sensing, our proposed on-demand sensing approach provides comparable high traffic estimation accuracy while significantly reducing the resource consumption. Based on the obtained results, using the on-demand sensing approach with 30% of cars as participants in the sensing activity, and a six-criteria matching approach yields a reduction of 73.8% in terms of network load and a reduction of 60.3% in terms of response time (when compared to the continuous sensing approach), while achieving a traffic estimation accuracy of 81.71%.