No abstract

No abstract

The transportation system has evolved into a complex cyber-physical system with the introduction of wireless communication and the emergence of connected travelers and connected automated vehicles. Such applications create an urgent need to develop high-fidelity transportation modeling tools that capture the mutual interaction of the communication and transportation systems. This paper addresses this need by developing a high-fidelity, large-scale dynamic and integrated traffic and direct cellullar vehicle-to-vehicle and vehicle-to-infrastructure (collectively known as V2X) modeling tool. The unique contributions of this work are (1) we developed a scalable implementation of the analytical communication model that captures packet movement at the millisecond level; (2) we coupled the communication and traffic simulation models in real-time to develop a fully integrated dynamic connected vehicle modeling tool; and (3) we developed scalable approaches that adjust the frequency of model coupling depending on the number of concurrent vehicles in the network. The proposed scalable modeling framework is demonstrated by running on the Los Angeles downtown network considering the morning peak hour traffic demand (145,000 vehicles), running faster than real-time on a regular personal computer (1.5 h to run 1.86 h of simulation time). Spatiotemporal estimates of packet delivery ratios for downtown Los Angeles are presented. This novel modeling framework provides a breakthrough in the development of urgently needed tools for large-scale testing of direct (C-V2X) enabled applications.

Software-defined networking (SDN), which has been successfully deployed in the management of complex data centers, has recently been incorporated into a myriad of 5G networks to intelligently manage a wide range of heterogeneous wireless devices, software systems, and wireless access technologies. Thus, the SDN control plane needs to communicate wirelessly with the wireless data plane either directly or indirectly. The uncertainties in the wireless SDN control plane (WCP) make its design challenging. Both WCP schemes (direct WCP, D-WCP, and indirect WCP, I-WCP) have been incorporated into recent 5G networks; however, a discussion of their design principles and their design limitations is missing. This paper introduces an overview of the WCP design (I-WCP and D-WCP) and discusses its intricacies by reviewing its deployment in recent 5G networks. Furthermore, to facilitate synthesizing a robust WCP, this paper proposes a generic WCP framework using deep reinforcement learning (DRL) principles and presents a roadmap for future research.