Schemes for multi-hop dissemination of non-safety information in VANETs

• Autores: Estrella María García Lozano
• Directores de la Tesis: María Celeste Campo Vázquez (dir. tes.)
• Lectura: En la Universidad Carlos III de Madrid ( España ) en 2016
• Idioma: inglés
• Tribunal Calificador de la Tesis: Juan-Carlos Cano Escribá (presid.), Florina Almenares Mendoza (secret.), José María Barceló Ordinas (voc.)
• Materias:
  • Ciencias tecnológicas
    • Tecnología de vehículos de motor
    • Tecnología de las telecomunicaciones
• Enlaces
  • Tesis en acceso abierto en: e-Archivo
• Resumen

  • Vehicular Ad Hoc Networks (VANETs) are a special case of Mobile Ad Hoc Network (MANET), whose nodes are vehicles and occasional fixed devices with communication capabilities. What makes them special is the limited range of possible movements of the mobile nodes (they can only travel on the existing roads or rails) and their high speed. The potential applications in this new type of network are almost endless. Researchers have typically classified them in four groups: active safety, public service support, improved driving and business/entertainment. The communication patterns that they require are varied, being information dissemination one of them. It is aimed at reaching a group of vehicles in an area that is larger than the reception range of a single node, so that a multi-hop broadcast is necessary. It can take multiple forms depending on the type of message. For example, a warning caused by a sudden brake requires a fast and reliable dissemination, whereas a blocked route announcement is tolerant to delays up to a few seconds and may miss some target without risking safety. The work in this PhD thesis is focused on this last type of use. The objective is to create schemes that would allow for the multi-hop dissemination of messages that do not have hard delay and delivery requirements (typically, any non-safety information). Our goals for this solution are four. First, we want it to be useful in roadways as well as inside cities. Vehicles movements and the occurrence of obstacles to the signal propagation are very different in both scenarios and so we need to adapt it to both. Second, we want it to be independent of infrastructure. The cost of deploying fixed units along every road and street is high and it may take a long time until there is global coverage. Our intention is that this solution can be used regardlessly of the deployment point. In addition, it must avoid the broadcast storm problem by reducing as much as possible the number of generated duplicates. Lastly, the scheme needs to cope with intermittent partitions in the vehicular network. Implementing a store-carry-forward mechanism that allows a message reach disconnected groups of vehicles inside the destination area rises the number of necessary duplicates. In order to achieve the aforementioned goals, we first study how typical infrastructure-less dissemination schemes from the state of the art in MANETs, plus a new specific one, apply to VANETs. According to their results in relation to a series of metrics, we learn that the distance-based scheme is the one that best meets our requirements. We select it to create an optimized scheme for the two existing scenarios-roadways and urban areas. Regarding the adaptation for roadways, we begin by optimizing the scheme so that its forwarding ratio is as close to the minimum as possible, and analyzing its average per-hop delay in a connected network (i.e., there is at least one feasible route between any two nodes in the network). Next, we study how to add a custom store-carry-forward mechanism that, with minimal additions, manages to overcome short-lived network partitions. We validate the addition and the complete scheme under different channel loads and in contrast with a well-known protocol aimed at the same type of traffic, DV-CAST. Our work on the version for urban scenarios parts from the assumption that we need to detect junctions and react accordingly in order to spread the dissemination in new directions and reach as many vehicles as possible. We create two different modifications of the basic distance-based scheme, each using a different method to detect intersections, and test them along with the basic one. This first step leads us to discovering that it is not necessary to detect intersections in order to achieve good results. Then, similarly to the process for the roadway scenario, we work on optimizing the scheme and creating a suitable store-carry-forward mechanism. We follow the same reasoning but this time we consider three different options for subsequent retransmissions. We test each version of the scheme throughly via simulations using real city maps and compare the results to those of the urban counterpart of DV-CAST, named UV-CAST. We use validated simulators as ns-2 and the Veins framework for testing realistically the different stages of our work. The performance of the resulting schemes meet our requirements to a high degree and so we consider that we have fulfilled our goals. In addition, the work done so far opens the door to new lines of research that are either the natural consequence or an application of our achievements.