Mechanisms, models, and tools for flexible protocol development and accurate network experimentation

In this thesis we argue that the growth of the Internet and the resulting variety of communication protocols and systems in general require new mechanisms, models, and tools to reduce the complexity and engineering effort of development and evaluation. We identify the following key challenges: Diversity and Customization of Communication Protocols: Typically, communication protocols are implemented from scratch making protocol development a challenging and time consuming task. We argue that the increasing number and variety of protocols in today’s Internet needs to be reflected in new mechanisms and tools that reduce this development complexity. Complexity of Protocol Evaluation: The heterogeneity of communication systems and variety of evaluation platforms require frequent reimplementations of communication protocols to be evaluated, rendering evaluation a complex and time consuming task. We believe, that the development cycle requires new mechanisms and tools that reduce the engineering effort in protocol evaluation. Lack of Realism and Accuracy in Network Simulation: While abstraction enables insight and controllability in network simulation, it abstracts from system properties, often limiting the realism of results. We argue that simulation requires new models to include system properties such as energy consumption and processing delay in the design and development process. Overall, these challenges are caused by the complexity and scale of distributed systems, such as the Internet and wireless sensor networks, the variety of applications and use cases, and the integration of heterogeneous systems ranging from tiny sensor nodes to super computers into today’s Internet. In the following we discuss each of these challenges in detail.