A Radio Resource Virtualization-Based RAT Selection Scheme in Heterogeneous Networks

Abstract

In the future heterogeneous wireless networks, heterogeneity of radio resources from different radio access technologies (RATs) still exists. The heterogeneity, especially for networks with the coexistence of non-orthogonal and orthogonal resources, makes the radio resources difficult to be uniformly measured, and thus hinders the efficient utilization of radio resources. To overcome this limitation, this letter firstly proposes a radio resource virtualization approach in heterogeneous networks. Based on the accumulated historical data of resource utilization information, heterogeneous radio resources are virtualized into normalized resources using deep learning method. Secondly, the consumption difference of virtualized resources under different situations of network load and user demand is modeled. Moreover, aiming at efficiently utilizing radio resources and reducing access blocking rate, a RAT selection scheme based on the radio resource virtualization is proposed. Through simulation, the validity of the proposed scheme is evaluated.