Multi-resource allocation in cloud radio access networks

Abstract

Computational resource allocation is a critical issue for the baseband unit (BBU) pool in a cloud radio access network (C-RAN). There are multiple resources in a BBU, including CPU, memory, disk, etc. The virtual machines (VMs) have diverse requirements along these resources to handle the baseband signal processing of corresponding remote radio units (RRUs). Consolidating VMs to BBUs based on single resource incurs over-allocation of the resources that are not explicitly allocated. Therefore, we study the multi-resource allocation problem in CRANs, which aims to minimize the number of active BBUs that are required to serve all users in the network. Since the RRU can be set to an idle state when its traffic is low, the number of VMs and their resource demands are all adjustable. We propose an efficient algorithm to solve this problem. This algorithm selects active RRUs and associates users with these RRUs in an iterative way. It adapts a heuristic for the multi-dimensional bin packing problem to assign VMs to BBUs. Our proposed method can significantly reduce the number of required active BBUs, while satisfying the VMs' demands for multiple computational resources. Simulation results demonstrate the effectiveness of our proposed algorithm.