A framework for reliable and efficient communications in vehicular networks

Vehicular ad hoc networks (VANETs) have emerged as a serious and promising candidate for providing ubiquitous communications, connecting vehicles to other vehicles traveling on the roads or vehicles to the Internet and other wide-area networks. This emerging communications platform can facilitate a number of vehicular applications - not only applications that are generic and already exist in other networks (e.g., Internet access, content distribution); but also applications that are new and specific to VANET. Due to the difficulties in seeking cooperation with the telecommunication infrastructure (i.e., those applications that require Vehicle-to-Infrastructure (V2I) communications), this PhD thesis will focus on applications that only require vehicle-to-vehicle (V2V) communications. Nonetheless, the work presented in this thesis can be used as a foundation and may be extended to the other type of VANET applications.;According to the Dedicated Short Range Communications (DSRC) standards, vehicular applications are categorized into two main types: safety and non-safety applications. Since these two types of applications serve different purposes and have distinct routing and end-to-end requirements, it is practically impossible to design a single communications framework (i.e., transport and network layer) which could support all applications. Two distinct communications frameworks are thus proposed in this thesis to facilitate the two application types.;The first part of the thesis focuses on the development of a new mobility model for traffic in urban areas and the comprehensive analysis of traffic pattern and connectivity characteristics of urban traffic. In addition to simulations study, an analytical framework that can predict key network connectivity characteristics is also derived. Unlike simulation studies, this framework leads to closed form results that provide valuable physical insight into the impact of key system parameters (such as road topology, traffic signal mechanism, speed limit, etc.) on network connectivity. These studies pave ways to the development of the communications frameworks for both types of vehicular applications (e.g., safety and non-safety applications).;A communications framework for safety-related vehicular applications is proposed and presented in the second part of the thesis. Since most of safety applications (such as Post Crash Notification (PCN) application) only rely on a single multicast transmission and hence, do not need an end-to-end control (i.e., transport layer), the main effort is put into designing the message dissemination protocol. In this thesis, we propose a broadcast routing protocol for urban VANETs, namely Urban Vehicular BroadCAST (UV-CAST) protocol. The proposed protocol is completely distributed and is able to support both disconnected and well-connected network regimes in urban scenarios with zero infrastructure support.;The last part of the thesis focuses on designing a communications framework for non-safety applications. Designing a Transport layer protocol in VANETs is a very challenging task as the end-to-end data transmission involves multi-hop wireless transmission; and the mobility and wireless channel conditions (such as fading, shadowing, multipath, etc.) exacerbate the well-known problems of using TCP over wireless links. In this part of the thesis, the two-way cross-layer control protocol (TCCP) is proposed. In addition to a fine-grained packet loss detection mechanism, the proposed solution addresses the end-to-end transmission problem at the network layer. However, unlike the conventional bottom-up approach, the network layer proposed in the TCCP scheme utilizes the transport layer information in making routing decisions through the top-down cross-layer design paradigm. The proposed TCCP scheme seems very promising for delay-tolerant applications and can support end-to-end communications in these applications for up to 5 hops.;In addition to the aforementioned techniques, fairness control mechanisms are also proposed for scenarios in which multiple data flows take place simultaneously. Due to the inherent fairness artifacts of CSMA/CA medium access mechanism, the proposed fairness control mechanism is designed to circumvent or minimize such artifacts; and based on the similar cross-layer concept, such the fairness problem is addressed at the network layer. In particular, a new routing algorithm based on a cooperative routing concept is proposed whereby vehicles cooperate in making a routing decision. In particular, each vehicle independently and locally selects the next packet forwarder in such a way that the total number of relay vehicles used to route all the packets is minimized. In addition to minimizing the MAC-layer artifacts, simulation study has shown that the proposed concept also enables fairness control to be managed locally at relay vehicles. By having vehicles (instead of wire- less channel) to partly handle bandwidth allocation among different data flows, the fairness issue can be addressed in a more efficient and flexible fashion. The resulting Multi-flow TCCP (MTCCP) protocol is extensively evaluated both in highway and urban scenarios and simulations have shown that in addition to fair allocation of resources, the proposed MTCCP protocol significantly improves the system performance in terms of goodput, end-to-end delay, and jitters.