Inter-vehicle communication networks have been identified as a promising solution to establish cooperative vehicular systems that overcome the current and future needs for increasing traffic safety and efficiency. To achieve their objectives, cooperative vehicular systems will be based on wireless communications between vehicles, and will have to deal with a highly dynamic mobility of the nodes, challenging radio propagation conditions, and stringent application requirements. As a prime example of upcoming ubiquitous networks that contribute to the vision of a thousand radios per person, inter-vehicle communication networks need to be designed to scale to high densities of radios while being robust and adaptive to varying radio propagation conditions but not making use of any centralized coordination entity. The goal of this thesis is an analysis and performance evaluation of the IEEE 802.11p communications technology based on which such inter-vehicle communication networks will be deployed. Since the performance is going to be affected by fast fading channel characteristics, employed signal processing algorithms, and medium access control, a solid understanding of all those three aspects is required.
To establish the required inter-disciplinary expertise, a tutorial style introduction to signal processing and wireless channel modeling is given in the beginning of this thesis. In a subsequent step, the different perspective and modeling approaches that exist in the corresponding but co-existing research communities are surveyed. Since the modeling approaches and the resulting simulators are incompatible to each other, a holistic approach is proposed that integrates all perspectives into one single evaluation framework. The developed simulator is validated against commercial transceivers in a controlled and emulated radio environment, and extended to include state of the art channel models that describe the time- and frequency-selective properties of the wireless channel adequately well. In order to reduce the computational efforts that are introduced by this new simulator, several GPU-based software architectures are evaluated with respect to their benefit on the simulator's runtime performance.
With the help of this new simulator, the coordination performance of inter-vehicle communication networks based on IEEE 802.11p is evaluated. The presented evaluation quantifies the level of incoordination that exists in such networks, indicates the reason of incoordination, and characterizes the negative impact of incoordination on the successful reception of transmitted packets. The obtained results indicate that (i) incoordination is inevitable due to the hidden terminal problem, (ii) fading does not influence the coordination performance significantly, and (iii) congestion of the network is the primary reason for a deterioration of the reception performance.
The identified scalability issue is discussed from the perspective of distributed congestion control mechanisms that adapt the transmission power and/or the packet generation rate of all vehicles. In contrast to the characterization of the coordination performance, the control problem is analyzed from the perspective of control theory with the objective to establish a set of design criteria which have to be met in order to obtain an effective solution. The microscopic considerations of this analysis show that cooperative detection of and cooperative as well as synchronized reaction to congested situations is essential. Unfortunately, the following survey of recent congestion control proposals shows that these insights are not widely recognized by the networking community.
In a last step, the inevitable incoordination that occurs in a decentralized network is addressed using the principle of successive interference cancellation (SIC) at receiving nodes. According to information theory, SIC allows to successfully decode two overlapping packet transmissions (one after the other) if they satisfy certain conditions with respect to their relative signal strength. According to the results obtained through a simulation-based assessment, these conditions are not fulfilled often enough in inter-vehicle communication networks, hence, the benefit of SIC is marginal in this context.