Processing Time Aware Resource Allocation in Software Defined RANs

Abstract

Computational elasticity concepts represent a quite promising area when it comes to virtualization and cloudification of radio access networks (C-RAN), especially within the context of 5G network architectures. The intention of such concepts is the provision of graceful degradation of quality of service (QoS) metrics under circumstances of scarcity of computational resources in virtualized networks. In this paper, we propose a new dynamic resource allocation strategy for the air interface that takes into account the computational resource utilization of virtualized network functions (VNFs) within the downlink transmission chain in terms of processing time requirements. The viability of the proposed concept is confirmed by an exemplary performance study addressing a typical deployment scenario within a software defined radio access network (SD-RAN).