Demo - MAC learning: Enabling automatic combination of elementary protocol components

Abstract

Cognition as a way to deal with the challenges of future wireless networks has been largely considered by the recent literature, with a main focus on physical layer adaptability and dynamic spectrum access. In this demo, we show how a simple cognition mechanism can be also applied at the MAC layer, by exploiting the emerging paradigm of programmable wireless cards. The idea is using the formal definition of simple MAC protocol components and platform-independent representation of channel events gathered from the wireless node, for emulating the behavior of protocols which are not currently running on the network, learning about their expected performance, and dynamically reconfiguring the wireless node. We demonstrate that programmable nodes, employing our cognition scheme, can find in a distributed way a con-conflicting schedule with other neighbor nodes and can switch from contention-based to scheduled-based protocols as a function of the network load.