A parameterized and optimized BBU pool virtualization power model for C-RAN architecture

Abstract

This paper introduces a cost effective scheme for a Cloud Radio Access Network (C-RAN) based on virtualization, which is a technology used to reduce Power Consumption (PC) in the Base Band Unit (BBU pool). The BBU's function is proposed to be as a software application running in a servers, which is called the virtual BBU (vBBU). To asses the proposal scheme, a power model of the BBU pool is proposed for C-RAN architecture. This provides a parametrized and optimized linear power model, which cover all the individual aspects in the C-RAN system relevant for the PC analysis in the BBU pool. Moreover, in this work a New Minimum Bin Slack (NMBS) algorithm is proposed, which combines with the Bin Packing Method (BPM) to minimize the number of vBBUs. The NMBS is aimed at matching the right amount of baseband processing load of Remote Radio Heads (RRHs) with load balancing as much as possible among all of the vBBUs, and switching off others. The baseband processing procedure for each vBBU is divided into N tasks, which can be processed by multiple vBBUs so as to offer system flexibility. The proposed architecture is based on integer linear programming for finding the minimum number of vBBUs. The simulation results show the NMBS effectively decreases the number of active vBBUs and decreases the PC in C-RAN when compared with other algorithms. In addition, this algorithm reduced the average PC in the BBU pool by 65%, 6%, and 3% less than for the Distributed Baseline (DB), the First Fit Decreasing (FFD) algorithm and the Heuristic Simulating Annealing (HSA) algorithm, respectively.