Exploiting Interference for Capacity Improvement in Software-Defined Vehicular Networks

Abstract

Vehicular ad hoc networks (VANETs), which are deployed along roads, make traffic systems safer and more efficient. The existing theoretical results on capacity scaling laws provide insights and guidance for designing and deploying VANETs. As a new paradigm of VANETs, software-defined vehicular ad hoc networks (SDVANETs) separate the data plane from the control plane. For many prospective applications, software-defined technology will be used in VANETs to achieve some general targets, such as network management. Therefore, a capacity analysis is critical and necessary for SDVANETs. In this paper, we propose a new fundamental framework named real vehicular wireless network model (RVWNM), which enables a more realistic capacity analysis in SDVANETs. We first introduce a Euclidean planar graph that can be constructed from any real map of an urban area and that represents the practical geometry structure of the urban area. Then, an interference relationship graph is abstracted from the Euclidean planar graph, which considers the transmission interference relations among the nodes in the network. Finally, we theoretically analyze the interference relationships in the interference relationship graph. A practical geometrical structure is used to calculate the asymptotic capacity of SDVANETs. To verify the feasibility of RVWNM, we calculate the asymptotic capacity of social-proximity urban networks. We also consider the social-proximity-based mobility of vehicles, and we derive asymptotic capacity bounds for sparse SDVANETs and constant bounds for high-density SDVANETs.