Decision tree based unsupervised learning to network selection in heterogeneous wireless networks

Network selection is one of key issues in the area of heterogeneous wireless networks. Actually there are lots of decision factors which are very useful for network selection. However, because the values of these decision factors belong to different types such as boolean, enumeration, discrete and continuous values, it is quite difficult to make use of these decision factors in traditional network selection approach. In this paper, network selection problem is formulated as an unsupervised learning problem. A decision tree based approach is then proposed to fully utilize the decision factors with different types to select network optimally.