Phase Timing Optimization for Smart Traffic Control Based on Fog Computing

The explosive growth of motor vehicles in urban areas has heavily burdened the existing transportation systems. Consequently, there are recently emerging smart transportation paradigms that aim to ease urban transportation pressure. Smart traffic signal control, which is considered as one of the breakthrough technologies in smart transportation paradigm, has ushered significant academic and industry efforts for its considerable benefits. The state-of-the-art proposals usually rely on a centralized infrastructure with powerful computing abilities to deal with a large amount of different traffic data. However, the centralized processing approach is often hindered by a long and even unbearable response latency, restricting its wide deployment and applications in the real world. To overcome this latency-related issue and achieve near-optimal traffic signal control in nearly real time, we have proposed a non-centralized approach as well as a fog computing-based architecture, and thus traffic data can be handled in a smarter way. To be specific, the traffic data are processed right at where it was generated, i.e., at the edge rather than at a centralized facility. In this paper, the phase timing of a single intersection is to be handled real-time by a local fog node with a genetic optimization algorithm, and the task for the regional optimization is offloaded to the centralized cloud and executed. The simulation experiments are conducted on the simulator to evaluate the performance of our proposal, and the results confirm that the proposed architecture and algorithm have significant improvement on the average duration time compared to existing approaches.