A Novel Influence Maximization-based Dynamic Forwarding Node Selection Scheme in SDVNs

Abstract

Software-Defined Vehicular Network (SDVN) has attracted considerable research attention as it can provide more efficient networking performance by decoupling the control plane from the data plane. However, in SDVNs, the flow table calculated based on the static graph can lead to inefficient routing. Besides, it is very challenging to select the relay nodes within the dynamic vehicular networks. To this end, in this paper, we propose a novel influence maximization-based dynamic forwarding node selection scheme (IM-DOS) for SDVNs. Influence maximization is a social computing technique for selecting the node of a graph for spreading information to reach as much nodes as possible. Here, we tackle the dynamic influential node selection problem of SDVNs by involving the time-constrained influence maximization (Trim) function from social society and learning-based prediction of link-duration, where this problem is NP-hard. We use the changing rate of distance, velocity, acceleration, location, direction, and graph information of the vehicle as input of Neural Networks to predict the links-duration. The link duration is also involved to guarantee that we can select the relay nodes with maximized impact and reduced overload. The extensive simulation demonstrated our scheme outperforms its counterparts, in terms of end-to-end delay and delivery ratio.