Simulating Data Flows of Very Large Scale Intelligent Transportation Systems

Vehicular Ad-hoc Network (VANET) simulations are integral to pave way for rapid adoption of Intelligent Transportation Systems (ITS) applications. Popular VANET simulators are built with high-fidelity network and mobility models to enable comprehensive evaluation. However, the computational complexity of high-fidelity models inhibits scalability. Further, the increasing infrastructure density of 5G and beyond increases the node count, thereby imposing further performance constraints. In this paper, we present a VANET simulation methodology called Disolv to enable large-scale analysis. The key decisions towards performance enhancement include simplifying mobility and network models, abstraction of packets to messages, utilization of a discrete-time instead of discrete-event paradigm for message exchanges, incorporation of data streams, and pre-computation of network links. The proposed methodology is implemented to evaluate the benefits through experiments. Results indicate that significant performance gains are observed with statistically irrelevant loss of precision. Further experiments are carried out to highlight the limitations of the Disolv approach. Finally, ITS applications suitable for evaluation with the Disolv approach are discussed.