Equipping automobiles with wireless communications and networking capabilities is becoming the frontier in the evolution to the next generation intelligent transportation systems (ITS). By means of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications, information generated by the vehicle-borne computer, vehicle control system, on-board sensors, or roadside infrastructure, can be effectively disseminated among vehicles/infrastructure in proximity or to vehicles/infrastructure multiple hops away, known as vehicular networks (VANETs), to enhance the situational awareness of vehicles and provide motorist/passengers with an information-rich travel environment.

Scaling law for throughput capacity and delay in wireless networks has been considered as one of the most fundamental issues, which characterizes the trend of throughput/delay behavior when the network size increases. The study of scaling laws can lead to a better understanding of intrinsic properties of wireless networks and theoretical guidance on network design and deployment. Moreover, the results could also be applied to predict network performance, especially for the large-scale vehicular networks. However, map-restricted mobility and spatio-temporal dynamics of vehicle density dramatically complicate scaling laws studies for VANETs. As an effort to lay a scientific foundation of vehicular networking, my thesis investigates capacity scaling laws for vehicular networks with and without infrastructure, respectively.

Firstly, the thesis studies scaling law of throughput capacity and end-to-end delay for a social-proximity vehicular network, where each vehicle has a restricted mobility region around a specific social spot and services are delivered in a store-carry-and-forward paradigm. It has been shown that although the throughput and delay may degrade in a high vehicle density area, it is still possible to achieve almost constant scaling for per vehicle throughput and end-to-end delay. Secondly, in addition to pure ad hoc vehicular networks, the thesis derives the capacity scaling laws for networks with wireless infrastructure, where services are delivered uniformly from infrastructure to all vehicles in the network. The V2V communication is also required to relay the downlink traffic to the vehicles outside the coverage of infrastructure. Three kinds of infrastructures have been considered, i.e., cellular base stations, wireless mesh backbones (a network of mesh nodes, including one mesh gateway), and roadside access points. The downlink capacity scaling is derived for each kind of infrastructure. Considering that the deployment/operation costs of different infrastructure are highly variable, the capacity-cost tradeoffs of different deployments are examined. The results from the thesis demonstrate the feasibility of deploying non-cellular infrastructure for supporting high-bandwidth vehicular applications. Thirdly, the fundamental impact of traffic signals at road intersection on drive-thru Internet access is particularly studied. The thesis analyzes the time-average throughput capacity of a typical vehicle driving through randomly deployed roadside Wi-Fi networks. Interestingly, we show a significant throughput gain for vehicles stopping at intersections due to red signals. The results provide a quick and efficient way of determining the Wi-Fi deployment scale according to required quality of services.

In summary, the analysis developed and the scaling laws derived in the thesis provide should be very useful for understanding the fundamental performance of vehicular networks.