In this thesis, we look at networking functionality as a set of services which can be composed dynamically. We argue that the dynamic composition of communication services can speed up design and experimentation with new protocol stacks. The reference Open Systems Interconnect (OSI) architecture of communication networks splits network functionality into layers, each providing a service to the layer above. The modularity provided by the OSI layered abstraction is further split in our approach by looking also at layers as a set of services.

In order to substantiate our claim, we propose a conceptual framework for quick and efficient prototyping and deployment of modular protocol stacks as a composition of communication services, and its reference implementation ProtoStack. The proposed framework consists of four components: the physical testbed, the module library, the declarative language and the workbench. The physical testbeds that ProtoStack, as the reference implementation, can support have to be based on hardware platforms which can host Contiki operating system. The module library used by ProtoStack is called Composeable Rime (CRime). The declarative language is based on the Resource Description Framework (RDF) and uses a custom built vocabulary. The workbench is a web based portal which allows easy, graphically supported configuration of a stack from the available CRime modules.

We show, through feedback collection from first time users, that the ProtoStack tool can significantly speed up prototyping and testing of new stacks and is friendly to novice and advanced users. The initial feedback shows that the tool can speed up design and prototyping of new protocol stack by at least a factor of 2. The cost of increased flexibility and prototyping speed of the protocol stack is paid in terms of increased memory footprint, processing speed and energy consumption. The CRime library used by ProtoStack has a 16% larger footprint, it takes 2.4 times longer to execute an open->send->recv->close sequence and consumes 1.6% more power in doing so. Even though with ProtoStack more resources are consumed by the node, the tradeoff in terms of prototyping speed seems to pay off.

The last part of the thesis discusses service oriented networks (SON) and cognitive networks (CN) and how the dynamic composition of communication services can be used to further research in this area. Particularly, we show on selected use cases how ProtoStack can be used to design and experiment with SONs and CN. The thesis ends with conclusions and a brief discussion identifying future work.