A Graphical Framework for Spectrum Modeling and Decision Making in Cognitive Radio Networks

There are many key problems of decision making related to spectrum occupancies in cognitive radio networks. It is known that there exist correlations of spectrum occupancies in time, space and frequency, which facilitates the decision making problems. A uniform framework, utilizing graphical models and tools, is proposed to integrate the spectrum correlations and decision making problems. Bayesian networks are used to model the probabilistic dependencies of spectrum occupancies. The statistical inference over the Bayesian network is carried out for spectrum sensing. Influence diagrams are used for cross-layer decision makings by integrating the inference result of the Bayesian network model of spectrum and the QoS requirement from upper layers. A simulated scenario of primary user network is used to generate the spectrum activities, based on which the proposed graphical framework is demonstrated to improve the performance of cognitive radio networks.