COSMOS: Composable, Systematic Cross Layer Design based on Multi-layer Optimization Strategies

We propose the Composable, Systematic cross-layer design and analysis approach based on Multi-layer Optimization Strategies (COSMOS). First, “Layering as Optimization Decomposition” serves as a base for a unified framework which provides a level ground for comparison among various cross-layer designs, as well as a unified view on “how to and how not to layer”. This mathematical theory provides guidelines that enable systematic design of network architectures. Second, in order to make feasible the development of useful software tools, we aim to enumerate alternative cross-layer architectures. Third, to make the theoretical guidelines useful in practice, theory needs to be translated into manageable cross-layer strategies. We propose the composable framework that enables flexible network evaluation and optimization. Composability is enabled by our unique technologies on applying ontology-based agent composition for multi-agent system models. The organization of the alternative cross-layer architectures can be naturally translated to an ontology knowledge base, embedded with rules and inference logic to guide cross-layer design. Finally, to analyze how cross-layer design may work with various security architectures, we impose architecture constraints in the mathematical formulation and “block” some cross-layer interactions. As a result, the designs out of such “reduced” design space will abide by the governing security architecture.