Deployment of roadside units based on partial mobility Information

Abstract

Vehicular ad-hoc Networks are expected to hit the streets soon. Given the automobiles role as a critical component in peoples lives, embedding software-based intelligence into them has the potential to drastically improve drivers quality of life. Leveraging wireless communication in vehicles has fascinated researchers since the 1980s. A few years later, as a consequence of the cellular revolution, voice services have become commonplace and ubiquitous, and the attention of researchers has shifted to wireless communication. In a vehicular network communication may happen in a direct and completely distributed basis where vehicles exchange messages without any support infrastructure. However, the ad hoc communication may become inefficient due to the high mobility of vehicles leading to very short contact times, possibly reducing the network throughput. Additionally, communication also suffers in sparse areas due to the lack of communicating pairs. Vehicular mobility also makes routing far complicated as we lack reliable means to infer the future position of the vehicles. Although the communication may take place in an ad-hoc basis, the research demonstrates that a minimum support infrastructure may largely improve the overall efficiency of the vehicular network. Nevertheless, the high costs of deploying the support infrastructure may delay the large scale adoption of such networks. Thus, several research efforts are towards maximizing the network performance through more efficient deployment strategies. Existent deployment proposals rely on two mobility paradigms: Initial deployment works are influenced by the cellular networks, and typically propose alternative strategies to identify the densest places within the road network. On the other hand, modern works assume full knowledge of vehicles trajectories, perhaps a not realistic assumption when we consider a real deployment. In this work we propose the deployment of infrastructure based on partial mobility information. Instead of assuming full knowledge of the vehicles trajectories, we assume full knowledge of the migration ratios between adjacent urban locations, a much more realistic assumption. We model the deployment of roadside units as a Probabilistic Maximum Coverage Problem (PMCP). PMCP allows us to project the vehicles flow in order to identify the expected number of vehicles reaching any given urban location. We use this information to infer the better locations for deploying the roadside units in order to maximize the number of trips experiencing at least one vehicle-to-infrastructure contact opportunity. Our deployment strategy is evaluated using three distinct scenarios with growing complexity: (a) theoretical road network; (b) real road network and synthetically generated traffic; (c) real road network and realistic traffic. Since traffic fluctuates, an architecture based only on stationary roadside units is unable to properly support the network operation all the time. Thus, we also investigate the benefits of dynamic infrastructure deployment strategies. Although traffic fluctuates over time, such fluctuation is somehow limited by the underlying (and almost) fixed road network. Thus, we propose a hybrid architecture based on stationary and mobile roadside units. Our investigation demonstrates that approximately 60% of the roadside units may be stationary. In order to address traffic fluctuations, the mobile roadside units must travel at an average speed of 5.2km/h, which demonstrates the feasibility of incorporating the mobile infrastructure into public transportation vehicles.