Management Strategies for Distributed Cross-Layer Reconfigurations in the Context of Cognitive, B3G Infrastructures

B3G (Beyond the 3rd Generation) wireless infrastructures are increasingly aligned with cognitive networking principles. Cognitive networks dispose mechanisms for dynamically selecting their configuration (algorithms and parameter values, at different layers of the protocol stack), through appropriate management functionality that takes into account the context of operation, profiles, goals and policies. This paper focuses on such management functionality by addressing a pertinent problem, dealing with "Distributed, Cross-Layer Reconfigurations" (DCLR). Our work contributes in four main areas. First, we formally define and solve a fully distributed problem version, which is very important for the management of a particular reconfigurable element, in a cognitive context. Second, we propose robust (stable, reliable), learning and adaptation, strategies for estimating (discovering) the performance potentials of alternate reconfigurations. Third, we give a computationally efficient solution to the problem of exploiting the performance potentials of reconfigurations, rating reconfigurations, and finally, selecting the best ones. Finally, we present results that expose the behavior of our schemes.