Connected vehicles are becoming ubiquitous with each passing year. Increase in mobilecomputing is proliferating the possible applications of connected vehicles. Many of theseapplications involve a continuous need for vehicles to connect to the communicationinfrastructure. This could result in congestion of the communication network. In thisstudy we evaluate a novel “dynamic grouping” methodology that combines vehicle-to7vehicle (V2V) and vehicle-to-infrastructure (V2I) communication schemes to make theoptimal use of the communication infrastructure. The methodology for dynamic groupingof instrumented vehicles is implemented in a realistic and well-calibrated microscopictraffic simulation test bed of the New Jersey Turnpike for the application of sensor datacollection. A reduction in communication infrastructure load of 66-91% can be achievedusing the dynamic grouping for systematic aggregation of vehicular information. Themaximum bandwidth usage is used as a measure to show that the name-address mappingis scalable. We show that the dynamic grouping methodology is very scalable withnegligible loss in data quality as compared to the scenario where each vehicle connects tothe communication infrastructure independently. The scalability is shown by generatingresponse surfaces for the load on communication channels for different marketpenetration and communication ranges. These response surfaces can also be useful inpredicting the channel load under future scenarios with increasing market penetration andpower of communication radios. The data quality is validated using reported speed andestimated travel times over the network. It is shown that on an average the error in speedis 5.5-8% albeit using far lesser bandwidth using the dynamic grouping approach.Similarly, travel time along different paths is shown to be within 5% during regularconditions and within 10% during non-recurrent congestion.
展开▼