13.07.2015 Views

System Architecture Design

System Architecture Design

System Architecture Design

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

pSHIELD<strong>System</strong> <strong>Architecture</strong> <strong>Design</strong>PUenvironments. Popularity and success of wireless networks and highly mobile nodes is currentlydominating in new development and research activities. In order to adapt to the rapidly and frequentlychanging network conditions under those circumstances, a more sophisticated interaction betweenprotocols than in a traditional layered architecture is desirable. The existing solutions are not able todynamically change which of these to use and how to use them, i.e., the adaptation, re-parameterizationand addition of cross-layer optimizations during runtime. Moreover, customization of optimizations inexisting frameworks is often cumbersome and complicated, if it can be done at all.Cross-layer architectures diverge from the existent network design approaches, where each layer of theprotocol stack operates independently and the data between the successive layers is exchanged in a verystrict and systematic manner. There are several advantages of a layered approach since modularity,robustness and ease of design are effortlessly achieved. The modularity that the layers provide allows forpotential arbitrary combination of protocols and the maintainability is being improved as new versions of aprotocol can be inserted without having to alter the rest of the network stack. However the properties ofthe different layers have substantial interdependencies and a modularized design may be suboptimal withregards to performance especially in satellite and mobile wireless environments, where thecommunication channels and traffic patterns are more unpredictable than in wired-line networks.There has been much talk about cross-layer design for wireless communication networks lately. It hasbeen argued repeatedly that layer boundaries, as specified in the layered architectures, are not suitablefor wireless communications and performance gains can be made by giving up strict layering to do crosslayerdesign [1], [2].This section discusses a communication methodology involving node cooperation which, whiledemonstrating a new opportunity created by wireless networks, significantly challenges the layeredarchitecture.Cross-layer design touches not just communications and networking, but is also intimately connected toconcepts related to communications architecture. Layered architectures have served to make the protocoldesign activity systematic and modular. Potential performance gains can always motivate a designer tonot follow the layered architectures and do cross-layer design. But cross-layer design cannot be seen asan end itself.This section presents both, a state-of-the-art (SoA) of the ongoing work and platforms over which newresearch can be built [3]-[20]. In this project we will discuss specific cross-layer design proposals and newalternatives. But mainly, we encourage a more holistic treatment of cross-layer design itself.We therefore propose a Cross-Layer pSHIELD <strong>System</strong> <strong>Architecture</strong> (CL-pSSA) with the followingproperties:• Signalling between an arbitrary amount of layers and system components• Extensibility of the architecture and adaptability of optimizations at runtime• High usability for cross-layer developers via an abstract description language for optimizationrulesThere also exist several cross-layer architectures facilitating signalling across all layers, i.e. any-to-anylayer signalling. For example, Cross-Layer Signalling Shortcuts (CLASS) enables direct signallingbetween all layers by message passing [8].PUD2.3.2Issue 5 Page 18 of 122

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!