There is a growing diversity in the range of emerging network applications such as social networking, online multiplayer games etc. and their expected network sup- port. Similarly, there is a growing diversity in the properties and capabilities of the underlying physical layer technologies ranging from copper wires, optical fibers, and wireless in the millimeter and micrometer wavelengths. While the traditional model of treating the network as a generic, best-effort data transport mechanism may have worked well for applications of the past such as FTP, email etc, emerging network applications having much complex requirements in terms of delay, jitter etc., have the potential to perform better in the presence of enhanced network support.

In general, there is a need for a mechanism that can map high level application requirements to underlying network capabilities and operator requirements, in a generic and extensible manner. This mapping would seamlessly enable existing applications to make optimal use of future technologies as well as allow future applications to make best use of current network technologies.

In this thesis, we propose to build a declarative framework that can reason over the requirements of applications, the current network context, operator policies, and appropriately configure the network to provide better network support for applications. The declarative specification of applications conveys information about the following : (i) Application constraints such as delay sensitivity, reliability etc. (ii) Type of data generated by the application and priorities of different messages (iii) Application adaptability to changing network conditions. In a sense, the declarative specifications formally express the intelligence that is typically embedded in applica- tions, and which can be used by the underlying network to adapt itself to achieve application goals under the current network conditions. In a way, we are attempting to formalize cross layer optimizations that have been typically handcrafted on a per application basis, in an effort to allow the network to reason and automatically apply the appropriate optimizations. Policies are employed to provide the network operator control over the reconfiguration process. Furthermore, we argue that by declaratively specifying applications, we may be able to identify similarities between application behaviors and apply optimizations developed for one application to the other. We use semantic web languages such as OWL, RDFS etc. for expressing application semantics and network policies, and exploit the languages’ inherent reasoning and conflict resolution capabilities. We will evaluate our system against other application specific optimizations based on simulator and testbed implementations.

In particular, the contributions of this thesis are (i) Developing a framework for using context and policies in declarative networked systems (ii) Runtime adaptation of network configuration based on application requirements and node/operator policy (iii) Formalize cross layer interactions as opposed to ad hoc optimizations (iv) Simu- lation and test bed implementations to validate and evaluate proposed approach. ]]>