Game Theory as a Tool for Modeling Cross-Layer Interactions

Modeling computer networks is a complex task, as their behavior depends from several variables. Focusing on a single communication device, ISO/OSI and TCP/IP layered protocol stacks provide interoperability and fast deployment of networking solutions, but they limit the control on the interaction among protocols operating at different layers. As a consequence, the need is emerging to develop appropriate models to capture and evaluate the interaction of protocols within a single communication device in order to underline such forms of "indirect" interaction - since they may lead to unforeseen performance degradations. The proposed work aims at using the game theory for capturing the interactions within the protocol stack of a single node, with the goal of allowing to determine the "steady state" or the operating point of the system in a given scenario. As a result, a scalable and modular framework is presented, that enables characterization and analysis of cross-layer interactions starting from the protocols' specifications. Finally, as an example of application, the model is applied to a single-hop IEEE 802.11 wireless network.