Epidemic information diffusion in realistic vehicular network mobility scenarios

Epidemic or gossip-based algorithms have been proposed for data dissemination in vehicular networks. Due to the unfeasibility of deploying large size vehicular networks, the performance evaluation of these algorithms is usually based on simulations. However, most literature works present experimental results based on hyper-simplified and non-realistic mobility scenarios or do not adequately describe the simulation setup. In this paper, we investigate the impact that road-network and vehicle density have on the performance of epidemic dissemination and the correlation between these factors and other simulation parameters, such as percentage of equipped vehicles, message expiration time, scheduling algorithm for data retransmission and number of circulating messages. We argue that, given the high sensitivity of this kind of approach to mobility scenarios, it would be useful to define a set of them to be used as a reference for evaluating different systems. Furthermore, from our simulations, we can state that, although epidemic diffusion does not guarantee data distribution to all network nodes, its simplicity and lack of infrastructure make it suitable for distributing real-time traffic or viability data in most mobility scenarios