Resource pooling via dynamic spectrum-level slicing across heterogeneous networks

Abstract

The performance gains from dynamic allocation of radio resources across multiple heterogeneous networks is studied. Through virtualization, the physical radio resources of the heterogeneous networks are first abstracted into a centralized pool of virtual radio resources. A dynamic spectrum-level slicing algorithm to share these radio resources across the different networks is then presented. This algorithm is responsive to changing user load and channel conditions. Simulation results show that for representative user arrival statistics, dynamic allocation of radio resources significantly lowers the percentage of dropped packets. In addition, they reveal that the triggers for dynamic allocation of resources across coexisting virtual networks occur every other time interval under the worst case traffic variation in the system (i.e., traffic varies every time interval). Our results suggest that performance benefits can be had even if dynamic spectrum-level slicing does not happen on time scales similar to that of the local resource schedulers residing in each virtual network.