This paper presents one promising application of Vehicle-Infrastructure Integration (VII) which is to measure arterial performance in real-time. It utilizes the information collected through the V-I communication in concert with those collected by conventional point-based detectors. In this paper, the average travel time is chosen as the major measure of effectiveness (MOE) of traffic conditions. A VII probe data (VPD) based model has been developed and customized for the latest VII probe message standard. In parallel, a point-based detection (PBD) based model with real-time inputs from inductive loop detectors and traffic signal controller and partial VII inputs has been developed. Lastly, a neural network based fusion model has been developed and trained by using the data generated by the simulation tool VISSIM. A six-intersection arterial testbed has been chosen to evaluate the developed models. For the simulation evaluation, the average absolute estimation error percentage for the PBD model is 13.9%. While the VPD model performs better when probe vehicle penetration rate is 5%. The fusion model improves the results by 35% under the same penetration rate. With the increase of penetration rate, the estimation results for the VPD model and the fusion model can be further improved. In conclusion, the simulation results show that the developed analytical models work pretty well and are able to produce accurate and reliable estimations along the testbed arterial. Moreover, the paper proves that the VII can be a powerful enabler for a wide range of ITS applications.
Samenvatting