Abstract-One of the most distinguishing features of vehicular ad hoc networks (VANETs) is the increased mobility of the nodes. This results in the existence of transient communication links, which degrade the performance of developed protocols. Established routes frequently become invalid, and existing communication flows are interrupted, incurring delay and additional overhead. In this paper, we aim to provide a metric to support the design of networks that can proactively adapt to a constantly changing topology. We present a method that produces a linklifetime-related metric capable of capturing the remaining time for which a link can be used for efficient communication. The metric is intended to be used to optimize route construction with respect to lifetime. We propose a cross-layer approach, which utilizes physical layer information, and formulate the relevant parameter estimation problem. Contrary to existing work, the method does not assume knowledge of the transmission power or the nodes' position and velocity vectors, or adoption of a specific mobility model, whereas the estimates go beyond describing the tendency of link quality. We achieve this by employing a unified model that accurately captures the effect of the radio propagation and the underlying structure of vehicle movement on the temporal dependence of the quality of a wireless mobile link. More specifically, the model takes into account the inherent nonlinearities arising and, most importantly, includes the minimum distance that will be achieved between two vehicles on the course of their movement, which is shown to play a crucial role in the link duration. We present an analytical framework, which quantifies the probability of correctly identifying the longest living link between two given links, based on the estimates. Utilization of the estimates is shown to lead to optimal performance under ideal channel conditions. The proposed scheme outperforms existing affinity-based schemes, achieving to opt for links that last up to 35% longer under the presence of shadow fading. Finally, we discuss the integration of the proposed estimation method in routing and demonstrate that our estimations can be beneficial, leading to the construction of routes that consistently last longer than routes that have been constructed based on the smoothed signal-to-noise ratio metric.Index Terms-Link lifetime estimation, link residual time (LRT), vehicular ad hoc networks (VANETs), vehicle-to-vehicle communications.