Cognitive learning is a very important part for cross-layer design in cognitive radio networks (CRNs).CRNs are required to take advantage of the known cross-layer parameters for learning environment and reconfiguring the network.This paper proposes a cross-layer learning scheme for CRN based on rough set

builds database of case events

knowledge base and rule matcher.This model solves the cross-layer learning in CRNs through combining data discretization

attribute reduction

value reduction and rule generation.By comparing the simulation results of typical testing data sets

a group of rough set algorithms are selected for the proposed model.The simulation results show that the set of algorithms can effectively solve accuracy and validity of knowledge extraction

rule generation for CRN cross-layer learning.The proposed model can be validly used in knowledge learning for CRNs.