Improving the Estimation of Residual Delay based Forwarding Method in Opportunistic Vehicular Networks

To accommodate the dynamic characteristics in opportunistic networks, ``carry-and-forward'' technique is utilized. In this paradigm, the next hop forwarding relay node is usually decided by inferring future contact opportunities. A practical method is to predict future encounter time using nodes' contact history. But limited by prediction accuracy, this method has only an uncertain effect on improving delay performance. To improve the efficiency of selected relay(s), this paper introduces a new metric used to deicide whether an intermediate node is appropriate to be selected. The metric is based on the estimation of Residual Expected Delay (RED), and combined with a threshold value to guard the fidelity of prediction based relay selection judgement. After then, a single-copy and a multi-copy two-hop forwarding algorithms are proposed. Finally, based on the underlying vehicular opportunistic network extracted from large-scale realistic vehicle trace records, we give an empirical evaluation and analysis of the proposed method. Results show the new metric is efficient in improving delivery ratio with a low delivery cost.